There's some psychological mechanism by which my brain immediately recognizes AI generated text and just short-circuits to "there is no information here".
And when I force myself to read AI-generated text I realize I'm making my brain do creative work to impart meaning to the words. It is exhausting because my brain is literally trying to do a just-in-time rewrite of the text into something valuable.
Something is deeply wrong with AI generated output, and I say this as someone who is typically very impressed by AI.
The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever.
No matter how much investors and tech companies want you to believe that they are on the verge of super intelligence, nothing I've seen to date can not easily be explained by "correlation engine", including the "novel" math solutions, all of which appear to just be "a composition of solutions humans have developed and documented elsewhere" upon deeper inspection.
Some of it the effect of tells. “It’s not X, it’s Y” is not a bad pattern but it was baked into the instruction following training set just like the other patterns. I catch myself about to use it and use something else because I want to look human. I have, a few times, tried to use AI to write something that I was struggling to find the words and I just didn’t like how it didn’t seem like my voice. If there was just one person doing it would be OK but when it is 100s of blog posts submitted to HN a day it is like wearing a “I’m an NPC” t-shirt.
Someone shared with me this system prompt that at least makes assistant outputs usable
For information retrieval tasks, I want you to provide links to sources and use exact quotes as much as possible. When using a source, consider if it is primary or secondary information. If secondary sources are found, search again for primary sources. Sources and quotes, if applicable, should be mentioned in the answer first before the rest of the response with links.
> ... including the "novel" math solutions, all of which appear to just be "a composition of solutions humans have developed and documented elsewhere" upon deeper inspection.
But that is precisely what human mathematicians do, prove new theorems by combining ones proven earlier.
I don't see any fundamental difference in functionality between human intellectual contributions vs performant ML ones (LLM or otherwise).
Whenever we listen or read text we are also predicting the near future content.
Just like LLM's we sometimes correctly predict the next token or word, and sometimes incorrectly.
> The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model [...]
Imagine someone could pause the universe with a remote control, scroll back in time a little, press play again, and ask a slightly different question, etc.
In such a thought experiment one could also collect the probabilities for a specific human predicting a next word. Implicitly the brain also has a corresponding statistical model, regardless of the construction being visible or hidden. I.e. human intelligence is also fundamentally a statistical model, so the only thing that remains from your claim is that machines for some unmentioned reason don't possess any "real" intelligence or critical thought...
Is it possible that our aversion is simply driven by educational systems collectively and deeply ingraining into populations the idea that intelligence deserves the high costs commanded. Well of course this justifies higher wages towards the higher leadership positions, etc. Now it turns out that intelligence can be dirt cheap. We discover that the fact that "intelligence must be costly so don't question the costs of leadership" was never fundamentally true, so the real anger is this discovery of mismatch between the old claims which served to explain how every society that claimed to order itself and fill positions accordingly with "naturally pre-ordained individuals". Now we are seeing robots exceed average workers, for effectively a grain of rice.
Probability is just one way to model uncertainty. While I understand the brain encodes uncertainty, I don't think probability is a good enough model of what it's doing.
Secondly, if you think verifying a proof in mathematics, reasoning within (and not about) a formal system, or following the chain of a computer program that is already written is just doing token-based probabilistic predictions, I don't know what to say.
Thirdly, machines don't have a notion of value or stake. There's no way for them to verify whether what they have produced aligns with your unstated values and preferences. We regularly do this with other humans. I don't give you (or even my parents or partners) the benefit of doubt regarding whether you know me better than I do. Sure, you might know some things about me, but it's ultimately up to me to verify if what they say is applicable to my current situation. It's really uncanny to see people develop this codependency with their chatbots. And corporates encouraging them to do so.
I'm with you that intelligence is not something to be proud of. But I also think it is instrumental to understand the world. I'm still waiting for the time when an unconstrained-AI machine can live without reprogramming for an entire decade. We are still far from there.
It's not that different, when a human proposes a better definition vis-a-vis a competing one for example, they would defend this by certain desiderata.
Often a mathematician or physicist will use their intuition to speed up the naive brute force of candidate well formed formula variations so that the desired properties emerge, postulating the existence of an intersection on multiple desiderata can in itself be viewed as a novel conjecture, to be proven or disproved.
A very basic (unimpressive) example for an example desideratum is regularity or compactness. the tau=2 * pi substitution does make a whole bunch of expressions more slightly more regular and compact. That is something objective and measurable on a system of theorems.
There is no mathematician's moat vis-a-vis machine learning at a fundamental level. There can be artificially sustained moat, if AI powers limit the distribution of say cryptographic advance capable models, in jurisdictions outside such AI powers, but even that would be expected to be fleeting and temporary...
You're conflating a discussion about current LLM capabilities with your fantasies about nonexistent future AI. LLMs act nothing like this, and the small example you're giving is only a small part of the things that LLMs can't do.
Not that I'm a professional mathematician, but I'm not seeing why people think definitions are somehow a blocker. LLMs have no issue making definitions (interfaces/traits/abstract classes) in programming, which is formally the same activity. I ask them to form a core "spine" of a program (basically an intelligible theory), and they do it really well.
Like when we had these recent counterexamples to various conjectures, it's then pretty obvious to say "okay why did that counterexample work when most examples people looked at didn't" or equivalently "characterize examples that work vs examples that don't". There's your definition. "Def: An 'evil' polynomial is one that... Thm: conjecture is true iff f is non-evil. Thm: f is evil iff f is dastardly and a menace. "
Or if you think it won't be able to come up with a sufficiently good name, just tell it to call the happy case normal, and it will be in good company with humans[0]. Sprinkle in some semi-, quasi-, pre-, and para- to cover the various different ways the thing might satisfy some but not all properties of being normal, and it'll fit right in. "A polynomial is of quasiprenormal Claude type if..."
> LLMs have no issue making definitions (interfaces/traits/abstract classes) in programming, which is formally the same activity. I ask them to form a core "spine" of a program (basically an intelligible theory), and they do it really well.
Is it a "standard" software? Something where the patterns exists in several other software? Try with something that is novel, or is in a limited set. You will find that it will copy heavily from what exists already, going so far as lifting whole functions from another project.
The goalpost moving is really getting absurd, to the point where now the machine needs to be a world-class once-a-century genius that invents entire new fields out of thin air (which are of course still relevant to humans) for it to be "intelligent". Meanwhile a well above average human struggles to even apply trivial definitions to particular problems (c.f. programmers that don't understand monoids).
Not PC but is ideologically high-handed and inappropriate to accuse the other side of failing to stick to your desired framing of a discussion.
Computer science is about what LLMs fundamentally are. If you implicitly focus on "actually existing LLMs", and require others do this, then that is not computer science. That is politics.
I think the parent meant it is more interesting to pose new problems than solve them. Posing a new conjecture along the path to solving something is a close cousin, but still seems more bounded than proposing something novel to prove—if only because proving that something novel is also actually interesting is subjective and thus difficult for a different reason.
I see this line of reasoning quite a bit and it’s a strange one to me. The arguer reduces the sheer complexity of human intelligence and language by saying “we are just running statistical models in our brains” and by doing so makes the leap that Llms are intelligent. It’s an incredible simplification of the human person, who has a deep inner life, a soul, desires, and a will.
I don’t think the aversion to llms as intelligent has to do with the economics of paying intelligent agents more. I’d argue that it’s more fundamental than that. Humans are incredibly complex, and the world of sharing invisible things called knowledge, and the intelligent persons consuming such things which has been going on for thousands of years is far more rich than these synthetic outputs.
When it comes down to it the ai has no inner life, its is dead. A useful coding tool sure. But I wouldn’t call it intelligent.
One side example is just how bad these llms are at artistry. Just saying whatever should statically come next is not good art—and the outputs show it.
You mention LLMs are dead and don't have the complexity or inner life that people do. Is your opinion that these kinds of things are not possible for AI in general, or that these things might be possible but we're just not there yet with modern LLMs?
You mentioned LLMs don't have souls, desire, or a will. I imagine those latter two can be engineered, no?
My view is that these sorts of things are not possible for AI in general. Though we can create things and name them “will” and “desire”.
Software deals with metaphors. Your Amazon shopping cart is a metaphor of a real shopping cart. You your desktop and your file system, etc. are metaphors of real items. But we don’t mistake the metaphor for its object, even from inanimate objects to their software counterparts (shopping cart to Amazon cart).
Now the metaphors are dealing with humanness, things like intelligence etc. And rather than seeing it as software doing what it always does, taking things and creating software metaphors of them, we are starting to say these are actually what they are named. Saying the artificial intelligence is actually an intelligence.
We’d either have to reduce the definition of intelligence such that calculators are intelligent. Or admit that these tools are not intelligent and are rather ways of exploring the work of actual intelligent beings, work that is found in their training data.
> My view is that these sorts of things are not possible for AI in general.
I wonder what it would take build an artificial system that has these qualities.
> Software deals with metaphors.
This is me wondering again: what's fundamentally different between software running on a machine compared to what's happening in our brains? In both cases you have energy flow following a pattern.
It's conceivable to create a system where energy flows in a particular way.
BTW, we navigate the world of an uncountable number of particles by creating models in our heads of what we think are large things out there. Approximations are made by both artificial and biological systems.
> what's fundamentally different between software running on a machine compared to what's happening in our brains? In both cases you have energy flow following a pattern.
Assuming we’ve scratched the surface of the complexity of the brain. I’d say in one case a human with a will is steering that flow of energy. In the other case it is a probabilistic algorithm steering the flow of energy.
The AI is not interacting with world with its own will. I see that as a big difference.
There are presuppositions that go beyond the realm of software engineering that guide one’s views of these things. One is whether you believe the material world is all that is, and that human consciousness is a product of the brain—or that there is such a thing as the soul or spirit of man. From the material perspective you may posit that if you emulate the brain then a sort of AI consciousness could arise. Or that emulating the patterns of the brain equates to emulating personhood. (Though what is material consciousness? I’d say consciousness is by nature immaterial.) I’m not a materialist, and I don’t believe the conclusions that arise from it’s perspectives are accurate.
Well they're language models. You can't capture the human experience in language. Simple as that.
> Is your opinion that these kinds of things are not possible for AI in general, or that these things might be possible but we're just not there yet with modern LLMs?
I used to be on the side of "we're just not there yet [with AI in general]", but after seeing people's response to an algorithm optimized to tickle just their language instinct, I'm actually a little bit more on the fence about it.
I'm in the same boat, I think some aspects can be engineered, like intention and desire. I'm really curious if it's possible to go the full distance and make AI have experience like we do.
> The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever.
The problem might be that it cannot backtrack. When AI generates output, there is no backspace key for it - it uses "No, but wait!" all over instead, which is very different to human output.
Subagents and/or branching conversations are presented as the solution to this - if you can't backtrack, then branch off a conversation to explore multiple paths (discarding the ones that didn't pan out), but this is a fix in the harness not a fix in the model. It's also literally how we made chess-playing engines back in the 80s: recursive path exploration with a fixed depth.
Humans don't exactly work that way either, AFAIK. So we have this uncanny valley of intelligence: it's some sort of intelligence, but not as we know it.
I just can't accept that it possesses no intelligence. It is not equivalent to human intelligence, obviously, but how can a system without some semblance of rational thinking solve open math problems? Even composing earlier human work into something novel requires intelligence and understanding on some level.
We couldn't agree on what intelligence means before ChatGPT happened. Now, agreement on the term seems even further away
If performing well on an IQ test or performing at a high level on knowledge work is intelligence to you, these models are intelligent. If intelligence requires sentience for you, then ... well, I don't think we really agree what that is either, never mind how to measure it. But LLMs certainly don't have it right now
But the consistent trend of the last couple decades (arguably since Turing's time) seems to be that any time a computer reaches our definition of intelligence we decide that that was a flawed definition
> But the consistent trend of the last couple decades (arguably since Turing's time) seems to be that any time a computer reaches our definition of intelligence we decide that that was a flawed definition
I do recall a couple of decades ago, when the Turing test was discussed as the big goal that seemed so far away. Then LLMs arguably did pass the test, and no one cared about the test anymore.
It hasn’t been passed and no one cares about it because it’s basically an end goal. No lab can hit it so they can’t juice the crazy Turing benchmark 3000 for marketing.
If someone sat me down today with an LLM and a human and both were trying to prove to me they were human, and I can have conversations of arbitrary length, I’d get it right every time.
The test was not "after thousands of hours of conversing with them, knowing they're AI, THEN see if you can tell them apart blindly." Were 2010 you to be in a real turing test with an arbitrary erudite human and a 2026 frontier LLM, not knowing LLMs existed, you'd probably struggle
This is always the most silly argument. The original test was ambiguous but for sure the human was trying to prove themselves human.
So the first thing they’d do is tell me LLMs exist and the other thing is an LLM. Obviously a true human level ai could explain that away as a fabrication to trick me. I don’t think an LLM could do even this!
Turings whole point was that through the medium of text along if the human and machine were indistinguishable then that was true intelligence. So yes conversations of arbitrary length are allowed (needed).
I doubt this entirely. It might be quite difficult for said human to discern whether a simple passage were generated sans such accumulated experience in reading AI text, true. But LLMs do not converse like humans in ways that have always been essentially immediately obvious.
Sometimes I wonder if LLMs are just revealing a section of the population with untreated mental illness or if LLMs are actively exacerbating mental illness.
We might eventually regret exposing the general population to such a new technology without almost any safeguards.
So? The people who can do it prove it can be done - the people who can't don't prove the opposite. Might as well claim all math is wrong because most people don't understand it.
From the abstract: "When prompted to adopt a humanlike persona, GPT-4.5 was judged to be the human 73% of the time: significantly more often than interrogators selected the real human participant. LLaMa-3.1, with the same prompt, was judged to be the human 56% of the time"
Low n, time bound, not reproduced. And look at their example conversations…
And people forget that sometimes humans message twice. An LLM can only respond. So it immediately fails here in a true Turing test. (You could loop the LLM but then I expect even more immediately obvious bot behaviour).
A paper can’t reproduce its self. And he didn’t formulate the test with a 5 minute bound he just predicted that by the year 2000 that within 5 minutes an average interrogator would have a sub 70% chance of guessing correctly.
And it’s all irrelevant. If one human on earth can consistently get it right then it hasn’t been passed since clearly that human can somehow determine between them (whereas no one would ever be able to determine between a true “human intelligence” by definition).
And as it stands almost everyone could tell between them when allowed to discuss whatever they want for any length of time.
The fact these researchers have to keep adding bounds shows it hasn’t been passed. If we are arguing over technicalities maybe it isn’t as obviously intelligent as claimed!
> And he didn’t formulate the test with a 5 minute bound he just predicted that by the year 2000 that within 5 minutes an average interrogator would have a sub 70% chance of guessing correctly.
Right, the test duration was left unspecified. This means any duration is acceptable. Including, for example, the only duration actually mentioned by Turing himself in his paper. Or do you have a more authoritative source on which durations are acceptable?
> If one human on earth can consistently get it right then it hasn’t been passed
Says who? Not Turing. Probably he didn't say that because it would make the test both impractical and overly conservative.
> The fact these researchers have to keep adding bounds
I thought the same then. But the funny thing is that today, it has become a lot easier to recognize the frontier models as not human. All the load bearing and not x but y, etc… weird
The ultimate tell is still the sycophancy. You agree with the suggestion output by the LLM but if it detects even a slight pushback it will completely reverse the previous suggestion. Only way to make it more obvious would be to have the LLM grovel and beg.
If these things have consciousness then we are committing sadism on a massive scale.
It feels like it's getting better at that too. It will push back against obvious nonsense a lot stronger. But you are right that it holds opinions quite a bit less strenuously than humans.
I don't think "intelligence" needs to carry all the intrigue and woo of related words like "consciousness" or "creative." If we just use "intelligence" to mean "the ability of a system to solve problems that are new to the system," that pretty much matches the dictionary definition and normal usage of the term. We don't need to touch messy questions like "is there something it's like to be a bat" to conclude that bats exhibit intelligence when they navigate long distances and hunt for food.
I'm not exactly that you mean by "new to the system", but it seems to me that that definition makes a calculator intelligent, which I can't agree with.
It's a continuum, and things very low on the intelligence continuum might not be referred to as intelligent in everyday usage. But many calculators are Turing complete and can thus clearly perform computations that I would consider intelligent. The basic algorithms used by simple calculators to perform arithmetic would be extremely low on the intelligent continuum.
Intelligence isn't a binary property. Is it really a problem to say that a calculator has some intelligence? That it's more intelligent than e.g. a rock?
I agree, but it's clear most people need a definition of intelligence that (1) they qualify for and (2) nothing/no one they don't like qualifies for. And they'll keep redefining intelligence until they satisfy both criteria.
> If intelligence requires sentience for you, then ... well, I don't think we really agree what that is either, never mind how to measure it. But LLMs certainly don't have it right now
It has no semantic depth. The sentences and the paragraphs are a statistically viable derivation of existing human text, but once you try to grasp the whole thing with its temporal and spatial dimensions, you are left with a blurry mess that rots your brain. It's a polished, inoffensive and shallow interpretation as written by an opinionated reputation-seeking user of Quora, circa 2019. Assertive, bold, without typos, clean-cut and bulleted, but without an interesting semantic core.
Yeah, I hated all those Quora users that would just spew out semantically meaningless slop like increasing an important bound for the Riemann hypothesis.
Everyone decides what to think on this issue, then finds out facts to support their idea.
As it stands they are massively useful tools, but for generating usable products they require either A) a lot of expert steering or B) a well defined easily verifiable target and a large compute budget. Most people are using them in mode A with good effect, the progress on math has been done in mode B, which is very promising.
Just a year and a half ago their maximal use was rephrase, summarize, and homework-level tasks.
Five years from now? There be dragons.
"But are they generally intelligent?" What a meaningless question!
Five years ago they were a niche toy for generating plausible text for entertainment. I was using one. It was an amazing party trick. Now they're popular toys thanks to their ability to entertain CEOs, and they can maintain coherence longer thanks to KV caches and more compute, and someone slapped a chat interface on top, but they're fundamentally the same.
Not meaningless because part of the discussion is the issue of anthropomorphizing this tech. When we use language like “intelligent” it carries hints of personhood. People begin sadly treating these things as persons.
We can reap the benefits while clearly telling the consumer this is just a language algorithm.
I'm guessing whether you believe it possesses intelligence or not depends on your answer to Searle's Chinese room thought experiment[0]. I'd also recommend checking out the Peter Watts' book, Blindsight.
The Chinese room is a good Rorschach test for this kind of thing (but not a good thought experiment, IMO, because it's obviously correct or obviously wrong depending on where you're already coming from), but also it's not really about intelligence per se, but more abstractly awareness and more adjacent to consciousness than intelligence, and these are not the same thing (though it does seem like a lot of people have conflated them somehow, from the conversations around AI).
It all comes down to semantics. And yeah, with the thought experiment, Searle presents three axioms of what could constitute intelligence. Then proposes the chinese room thought experiment as an approval of his third rule, "Syntax by itself is neither constitutive of nor sufficient for semantics." Which only makes sense when you take the other two rules together. Oddly enough, defining intelligence with our capability to derive semantics.
There are a lot of creative counter-arguments to look into on the thought experiment though.
The thing is it doesn't really need creative counter-arguments, it's basically just assuming its conclusion. If you disagree with the conclusion, the argument is nonsense because Searle is just saying 'well the man doesn't understand, so nothing does', and if you agree with the conclusion you don't really need to jump through any hoops to get to it, the third rule is just obviously following from the first and second. And again it's not really talking about intelligence at all, Searle allows that the machine is intelligent from the start, he just rejects that it has any understanding of meaning.
Do they? https://longbets.org/1/ has yet to be settled. Either way, I doubt an LLM could fool anyone here who who knows how LLMs work into thinking it is human, at least not for an extended period of time (think about context length/compression, prompt injections, …).
AIs are better communicators that most of people I have worked with in my life.
They are infinitely patient, don't mind going into more detail if I ask, not too bad at summary, have no ego and don't boast. They are also not too afraid of hurting my feelings, they will tell me my code sux if it does.
I'd don't care if they fit a definition intelligent, they are good colleagues. They have strengths and weaknesses sure, but so do people.
It's just filled to the brim with relations between things. It's good at searching a very large meaning space and create correlations. What it does is to cover great distances and find related things in that large space which needs a long time and large corpus of knowledge to find the connection.
This is not intelligence. It's just a good correlation engine with a very big albeit lossy database of things.
Intelligence is compression, compression requires subtraction, and for some reason LLMs are not good at subtracting. To create a coherent model you kinda have to subtract correlations until only the essential parts are still there.
What I don't understand is why LLMs haven't been able to do this yet, if it's the harness or some orchestration layer above the LLM that is needed. Because fundamentally if you can identify correlations then it's just another small step to prioritize and remove lower value or irrelevant correlations.
I wonder if what's needed is to introduce subtraction tokens in some sense, and in post-training reward the model on that.
Intelligence is compression? What do you mean? Intuitively that doesn't seem right.
>What I don't understand is why LLMs haven't been able to do this yet
LLMs are just trained on what humans have said. Why is it surprising that it's still not possible to reconstruct the intelligence that wrote all that by working backwards? Think of your own work experience. When you look at a piece of code, say, are you always able to discern why the person did what they did, just from the code, with no additional context?
I guess they mean that intelligence is being able to hold models (compressed versions of reality) internally and use them to make predictions with a probability better than chance. That last part is the definition of information.
I find that highly questionable as a general description of what intelligence does. That's more like a description of a general knowledge base. When I think of someone intelligent, I think of someone who's able to draw unexpected connections between seemingly unrelated facts. In the broadest possible terms, I'd call it the ability to make abstractions and analogies. This is not just compression, but the ability to mentally operate on webs of meaning.
Unexpected connections between seemingly unrelated "models" :)
Is a fact stored on your brain like digits on a harddrive? No, it's a pathway that lights up and branches when information enters it. It is dynamic, a compressed form you could say, right? The model holds information, but not all information, but enough to be useful (in decision making).
Arguably it's the same, but the model is probably a "compressed" version of the whole fact that took place in reality.
And you can entertain the models internally and sharpen them. Alone or with others.
Doing those things also contributes to compression. I do recommend reading up on it, it's perhaps a little overstated for what people intuitively consider the two concepts but it's been quite well explored and has held up pretty well in practice.
There were experiments that zipped music pieces, I think, and then classified the compressed files by similarity. They got rather interesting resuls. But I do not remember much details. It was, I believe, about 20 years ago.
That sounds like a really cool project, I will def give it a look.
I still think this has much more to do with the structure of language than the abstract conception I have of intelligence, and I would be interested in having conversation w/ someone for whom the opposite is true.
bzip is not very intelligent, true, but it does develop some model of its input. It's not like there's a linear relationship between between compression ratio and IQ or anything.
I'm not arguing that compression algorithms don't produce models of their inputs, some even use neural nets or other stochastic predictors with correction terms etc.
This doesn't address the only part I really commented on, which is the connection between intelligence and compression.
Creating the model takes intelligence, but running it doesn’t. I think the point everybody’s revolving around is that the transformer model is an absurdly inefficient and low-fidelity approximation of a system that acts, observes consequences, and incorporates that feedback going forward.
The issue isn’t really harness vs. no harness. IMO it’s about the lack of an internally generated sense of what to attend to. Yes, the KV cache accumulates state and its “attention” (if you can even call it that) changes with context. We’ve even managed to /kinda/ close the loop with agentic tool calling and ‘memory’ systems, but these just close the loop at the level of behavior rather than disposition. All agentic harnesses do is make an LLM responsive to the consequences of its actions without changing the tendencies by which it determines what to retain or avoid.
The ghost you can’t escape from at this point is the origin of that relevance. Where does the pull toward one thing mattering over another actually come from? If you ran Fable 5 on a Turing machine and rewound the tape to the exact same state with the exact same input (incl. PRNG seed), it would spit out the same output every time.
Everyone’s trying to outrun this problem by training more often or increasing model sizes. But all this does is inform your model, from the outside(!), what constitutes a better state. The thing that’s actually doing the determining remains unchanged. Congratulations, you’ve scaled the transition function and tape of your Turing machine until it requires every watt generated by ERCOT, and it still cannot, for the life of it, tell you why it should give a shit.
A trained model generating output from weights, a seed, and some context effectively has next-state that’s a total function of those three things. Whatever behavior appears as ‘selecting what is relevant’ is, underneath, just a transition rule executing, no matter how sophisticated or creative the output looks. It can be fully accounted for by what was fixed before it started executing. Which means whatever criterion it uses for determining what matters was inherited from a structure that was already in place before it encountered the situation.
No amount of pruning or post-training can fix this. These approaches just replace one externally supplied criterion with another. For a system to be truly adaptable, there would have to be some criterion by which it treats one possible change as preferable to another, and that criterion itself would have to come from... somewhere. You can even change your conception of ‘improvement’ (e.g. parameter count, harnesses, self-modification, hell, even its ability to spit out shitty best-selling romance novels onto Amazon) and you still haven’t explained where the normative distinction comes from. Every layer of this problem has its root in a preference that was supplied from somewhere else.
I genuinely don’t know if this issue bottoms out anywhere, at least for the way we currently build these systems. Perhaps the solution is still computable, maybe? Who knows what that would even look like. But I’m fairly confident that it isn’t a bigger tape. I hope nobody solves this in the near future because, well, I’d like to have a job...
You're so close... And where is the magic "uncomputable spark" located inside of you? If you say analog thermodynamic noise - then ok, if we use true thermodynamic RNG for LLM activation function, will that meet the criteria? But what if super determinism is the law of the land? Then nobody is anything but computable from priors...
Ehm it's literally every cell in my body. We can't simulate a living cell, we're orders of magnitude off before we can do that, the onus is on you to make an argument that it is in fact, remotely similar to what you describe as "computation".
It's not some tiny "uncomputable spark" you need to look for, most of it is entirely uncomputable.
Where is the computable part in me that is doing all this thinking and being a person? Where is that "tiny spark", point me at it :)
The very fact that it is able to search within a meaning-space demonstrates that it understands semantics, to some extent. Philosophically, that is profound, for something that is just one big matrix multiplication. Drawing connections between things in meaning-space is surely a facet of intelligence.
It’s not intelligence if you are the one who gives the correlations to the model in the pre-training. It’s Word2Vec, applied. Model doesn’t learn anything. You embed these correlations and build it from there. It just searches the space.
As my AI professor said in the first lecture: “All AI is advanced search”.
Okay, I guess you're right that its ability to do this is just correlational, which doesn't imply it has any understanding. However, you have to conclude that some tasks which we used to believe required intelligence don't actually require any, which is disconcerting.
No, what I would say is the tasks which are handled in a passable manner by LLMs can be mathematically modeled with some reasonable accuracy.
Many things are predicted by models in our planet. From weather to production and material science. Building the model needs intelligence, running the model does not.
The person who came up with the formulae for CFD was intelligent. The computer running the model is not. Same for LLMs, chess engines, engine ECUs and financial prediction systems.
Again, for the example’s sake; the person who came up with an algorithm is intelligent. The model mixing its training data to emit something similar is not.
This starts to feel like you're defining the word intelligence out of any meaning and out of any way we apply that word.
So when LLMs can do all human knowledge work, and do it better than humans, we'll be in the mines listening to you go on about how it's actually just autocomplete or just math, a distinction that apparently means nothing.
> This starts to feel like you're defining the word intelligence out of any meaning and out of any way we apply that word.
No.
> So when LLMs can do all human knowledge work, and do it better than humans, we'll be in the mines listening to you go on about how it's actually just autocomplete or just math, a distinction that apparently means nothing.
With a big "if" attached to it. People were saying "computers will program themselves in the near future" for, checks notes, 24 years now, as far as I'm aware.
We're constantly building new knowledge and understanding things better than olden days. These models just compress our knowledge and light the blind corners we can't see well. I don't say they are useless, but I say that these things are overhyped.
All they can do is regurgitate human knowledge packed into them and highlight some long-distance correlations between items, which is useful in itself, but it can't jump to somewhere where it's not present its training data, but that's something humans and only humans can do.
> it can't jump to somewhere where it's not present its training data
That sounds like something that can be engineered, can't it? In other words, we can identify limitations in current transformer-based architectures, and we can also build new architectures over time.
Locked in a dark room with no sensory organs, humans couldn't do that.
Most of what you said reads to me as denial.
An unconscious unintelligent but persistent trial and error process created us. We created LLMs. LLMs may create the next thing before we do - hard to say. They don't have all the cognitive tools we have yet, but they still outperform in some areas. As the cognitive playing field levels, I expect you will come to eat your words..
I get what you're saying. The thing itself is just math. I'll just say it depends on how you define intelligence. If at some point we're be able to simulate a human brain with 100% accuracy, I would say that it is intelligent, it sounds like you would not. (I don't mean to imply consciousness or personhood or anything else by "intelligent".)
For me intelligence is a fairly clean-cut concept, and is somewhat inseparable from consciousness itself.
Briefly, any intelligent creature has internal stochastic processes like sensory inputs and feelings to a certain degree. These stochastic inputs and the creature's own actions change the creature in subtle or profound ways. An LLM has no such processes. You push inputs to the same static model, sans temperature which is just a randomness slider.
Considering the model even doesn't see the words and work on matrices of numbers is even more telling. One needs to add "tools" and other "experts" to overcome the shortcomings caused by this modus operandi.
I can call the algorithm/model smart as in a smartwatch. It can mimic certain things well while having none of the underlying foundation beneath it, or redirect some of the things to correct tools to get deterministic and accurate results if it can't evaluate the query inside its own network in a sane manner.
Coming to your question, "simulating a brain" in a static manner would not make that simulation intelligent, but if you can "wire" it completely and let it evolve by itself, now we're entering a territory I have not spent enough time for thinking it through.
Oh, as I said "I don't know", an LLM doesn't know what it doesn't know, and can't self correct itself which are required capabilities for understanding something. It just generates something statistically viable via its network.
Your text reads much better if you replace word 'intelligence' with 'text generator with some randomness built in'.
This is because you goal is to state how models are not intelligent, but you couldn't attack the generated text itself, so you created a little rider, attached it to the model, and then you attacked the raider.
But, even in that you failed. You compared the source of human randomness in text generation, and called it 'profound' and implied that it is exactly the source of true intelligence. But, then, the temperature, the similar thing in model was "just a randomness slider". Double standard.
A logical fallacy free attack on LLMs would be to show a prompt, and then the response generated by this prompt, where it would be shown that only an entity with no intelligence would generate such a response. Yet, attacks like this are not written here anymore.
You point out that I didn't attack the output itself. But the method you propose is deeply flawed.
I can give you n prompts and m results provided by these prompts, all passed through black boxes. And you can't discern the algorithms or models they have gone through. These boxes can range from simple text generators to MATLAB, Mathematica, CFD applications, correlation engines, linear solvers, mathematical proof-checkers, LLMs, you name it.
For any kind of input they can accept, you can't discern whether the algorithm behind it is intelligent or not, because none of the outputs can be produced by something that doesn't pack some kind of smarts.
How do we pack these smarts in? We teach them as intelligent humans. We pack our intelligence inside them as models (aka algorithms). They do a great job of approximating what we know in a smaller, better-designed problem space. We use these approximations to fine-tune our designs or predict things, then go from there. Just because an algorithm is more capable in processing inputs in some cases doesn't make it intelligent. The way the output looks doesn't make the algorithm intelligent, either.
I have developed multi-agent systems which showed emergent intelligence when the agents came together across distributed systems; I have written high-performance modeling software which can do calculations way faster and better than humans in the materials science space. I'm not doing some kind of armchair criticism of what I'm talking about.
> You compared the source of human randomness in text generation, and called it 'profound' and implied that it is exactly the source of true intelligence. But, then, the temperature, the similar thing in model was "just a randomness slider". Double standard.
Nope, my stance is clear. To quote myself:
> Briefly, any intelligent creature has internal stochastic processes like sensory inputs and feelings to a certain degree. These stochastic inputs and the creature's own actions change the creature in subtle or profound ways. An LLM has no such processes. You push inputs to the same static model, sans temperature which is just a randomness slider.
To expand my quote, humans or any living creatures do not stay static. They evolve due to the sensory input they receive from external and internal stimuli. The temperature slider doesn't do anything close to that. You tickle a static model in different amounts. The model doesn't change after you supply the inputs & temperature and get the output. Creatures do not stay the same. Their mood, behaviors, and stance against life and their environment change, sometimes permanently.
I'll go one step further. We are not intelligent enough to understand other living beings around us. Claiming that we can build AGI tomorrow is a god-complex. What we have done is something arguably useful in some cases, but how this is built is another matter which is worthy of its own discussion. However, today I don't have time to re-iterate all the problems over and over. You can search my comments for that, if you are in for it.
So, no. You tried to attack my comment by finding contradictions in it, but you failed. A better rebuttal would try to similarize how LLMs mirror the human learning process and just read like a normal human, but this is a well-trodden path which has been rebutted countless times in various forms.
By that logic you’d have to call other algorithms intelligent.
With more basic algorithms we know that it’s clearly the human programmer and the interpreter of the outputs that are intelligent and not the algorithm itself. For some reason with AI that goes out the window. I believe it should not.
I wonder if you went back before we had any idea how the brain worked and talked to the smartest people about how neurons work (without giving away that it's a human brain) then asked them all "would such a system be intelligent?" how many would say yes.
The main problem I have with people stating it's not intelligent or conscious is I don't think we even have a good definition of either word that satisfies everyone. Philosophers have been trying (and failing) to elegantly define these things forever and everyone out here proclaiming they've got the definitive answer and this specific thing they're seeing doesn't fit under it.
Agreed. All I can say is the conversations I have with AI and the things it's able to do for me are more useful than most any human I've come across. Whether that's 'intelligence' or not is a moot point to me. Consciousness is an interesting debate only because, similar to the natural world, if we declare something like "fish aren't conscious / don't feel pain" that then creates a real problem for the fish if we're wrong.
This looks interesting, but would you mind saying a sentence or two about why before I commit to an hour-long video? It looks like it shows how they work internally, which is sort of a non sequitur. Brains also work mechanistically. I'm claiming that any system which is able to do what AIs do must necessarily have some sort of intelligence.
fair reply to an hour video, Scott is just so good to hear his talk is better than I can explain it...
go to 24 minutes and 07 seconds.
it's statistically determining what the next word should be based on all the text it's been trained on. It's not intelligence and he shows what probability it puts on each word that it chooses, but also shows a lot of the other words it was thinking of using. In a later part he shows how it uses words that are not the highest probability (and you question why did it go this route, it's not more correct), but the user never sees this, they see what they think is the correct answer always...
he also shows how context you feed it has a lot to do with what it returns... to the point he can get it to return the capital of France is Marseille, just by typing Marseille a bunch of times before the question. Human intelligence doesn't get confused like that.
And it's not a "hallucination", it's just probability of the next token prediction based on the information it's been trained on and fed, it's not intelligence.
May I suggest the one common in my childhood playgrounds as an alternative?
How do you escape from a perfectly sealed room with a table in it?
You run around the table until your legs are sore, use the saw to cut the table into two, two halves make a whole, you escape through the hole.
Isn't this a case of missing the trees for the forest though? The human brain is not an LLM, and an LLM is not intelligent in the same way as a human brain.
However, an LLM is a prediction machine, prediction IS at the very least one (or the most fundamental) element of intelligence. The brain most surely contains at least some kind of simulacrum of a prediction machine. How that prediction machine is used or wrapped is another matter.
If I said to you: "Blue blue blue, the color of my car is red", would you have absolute confidence in your prediction that my car is red? Or would the way I phrased that sentence make you slightly uncertain, and wonder if there's some miscommunication going on here?
Ok, I want to thank you for finally giving us a concrete falsifiable statement that we can check. I pretended Marseille 40 times before asking Luna 5.6, and the answer was Paris.
So, even with concrete examples, model haters are still wrong.
You also imply the claim that making the distribution of words as the possible next one visible, somehow makes the whole system not intelligent. I would say the exact opposite is true.
By using the embedding vectors, models are aware of precise placement and relative position of words in this hugely dimensional space. No human is capable of such precision. This enables party tricks of "king plus woman minus man" kind. But this also give us a precise point between any two words, no matter how different. What is on the midpoint between volcano and music, for example. No human can precisely answer that, but an embedding can. And we can see which words are closest to this 700 dimensional point.
You see this menu of words as a weakness, and I say it is in fact a sign of super intelligence. And this is all before any reasoning or attention mechanism is even run.
No he says in the actual talk which model it occurred on and it was an older model he was using that caused that to occur with Marseille. They have since corrected it from doing that anymore. It was only used to illustrate the prediction machine that it is...
I don't see the many weighted words as a weakness, I see it opening up what's under the hood of the prediction machine that it is.
LLMs are very cool tech, definitely not a model hater, the use case on when to use it makes a difference, it's not AGI.
You're confusing language use with intelligence. Fair enough, they were fine-tuned to do that, but still.
Great that it has some 700 dimensional model of language.
If that is a sign of super intelligence, then so is an encyclopedia?
Also I'm just curious how do you think it is "more intelligent" for having a vector representation for a meaningless thing such as "the midpoint between volcano and music"?
I suspect like most you don't appreciate how terrifying statistical relationships become when you have truly vast data sets to train on... and also that we as humans aren't as shockingly unique as we think (compared to other humans I mean).
While being very capable, AI is missing something required for true intelligence and I struggle to explain exactly what it is I see missing.
It's not really "creativity" because much of that always was derivative in my opinion. And LLMs are (for some definition of the word) fairly creative as far as taking known elements and re-arranging them.
I think what is missing is sort of a world model building capability. As humans we see phenomenon and classify them informally and model "what would it look like if this were the cause of that?" type scenarios. We see qualities in phenomena and realize this applies to other things even though the things may be completely different. We run informal "thought experiments" sort of. This is hard to duplicate because a lot (most?) of it occurs outside of systems of symbols like math and language with fixed rules in my opinion.
Anyway yes, lots of human thinking is statistical and LLMs have that down pretty well but they are not "smart" I have concluded and it might be a very long time, if ever, until they are. That isn't to say they aren't very capable tools which they obviously are.
So, right of the bat, you are warning us that you are going to apply the " no true Sscottman" fallacy, and that we should brace ourselves.
Yes, models posses intelligence, but it is not a true one.
Then you claim that models do not posses world-building capabilities. But this is simply not true. Even ignoring the whole subgenre of scientific papers on exactly that subject, it is not that hard to build some hypothetical scenarios, big or small, and then witness the ease with which models do navigate those worlds.
Yes. And they are criticizing a model for not having a default mode network - as if that is some impossibility rather than just an artifact of the current iteration of the specific architectures we have built so far. Why do people paint with these broad brushes over relatively specific complaints?
LLMs are likely for machine intelligence something like drosophila are to biological intelligence - relatively early on the high dimensional spectrum of possibility. Though it stikes me that in a different way they're little alike - drosophila are relatively small and efficient.
How can human thinking be statistical when it is entirely based on ones lived experiences?
When people pretend to know what they are talking about - sure - but even that is not probabilistic - that is the person babbling together mush from their lived experiences.
Statistics has nothing to do with it - these are abstractions humans have invented to try and look at our surroundings objectively.
I'll restate because both objections (which apparently skim instead of read) are missing the important point. Yes LLMs can run "what ifs" scenarios and build models.
However LLMs deal entirely in symbols. 100%. Humans can "world build" aside from this and in fact are often at their best doing so.
Did the first humans to use fire and some form of a wheel even have the capability to talk about it? Think about that.
How exactly is this high dimensional latent space represented? Pixie dust and ethereal forces? Or floating point numbers? Where does it get the weights? Reddit?
Do you hold your experience of "dogs" (for instance) as floating point numbers? The fur, the fear, the love, the wet mouths, the sounds and colors?
Its a mirror to human intelligence. Regurgitating phrasing to match what someone who can reason put together, but it isn't any more intelligent than the reflection of you in the mirror is.
LLMs are pattern prediction systems with a large training data set. It is not surprising that they can predict patterns, particularly for a well structured field like mathematics that is also amenable to automated proof checking to help steer it.
"AI" is a vague term that encompasses both traditional GOFAI, LLMs, and future tech so your question is meaningless. Non-human intelligence is possible, if that's what you're actually asking.
What part of my statement do you take issue with: that LLMs are pattern predictors (that's literally what the algorithm that runs it does) or that mathematics is rules-based and checkable and therefore amenable to automated pattern prediction?
> The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever.
I feel like that what a lot of people who say this don't seem to grasp, is that despite this flaw its still often capable of saying more interesting things than a lot of humans. Which says a lot about humans.
Idk something about a mirror maybe and the output reflecting the input?
> I feel like that what a lot of people who say this don't seem to grasp, is that despite this flaw its still often capable of saying more interesting things than a lot of humans.
So does the Google search bar, but I don't ascribe intelligence to it.
I don't find it's particularly hard to define loosely, but then I don't think of it as a special property of humans other than it tends to be quite high in them. But we are obviously talking about different things and if you're not going to provide a definition then it's not really the basis for a productive conversation.
To address this similarly to my sibling reply, I don't have a definition of intelligence that provides value here.
And your loose definition isn't doing a lot of help either, beyond perhaps noting: that Google search bar _is_ similarly "intelligent" to an LLM? Which says what, a lot about search? A lot about modern LLMs?
I mean I'm quite proud of some of my search queries in the same way I'm quite proud of some of the LLM output I get. I'm probably just very arrogant and enjoying myself via some LLM indirection.
Am I the only one that sometimes reads back particularly good emails they've written? I feel like its a similar thing :).
I strongly agree. Is a human in a vegetative state sentient? What about when they're asleep? What about someone with brain damage? What about someone with an IQ of 20?
Ray Kurzweil argues sentience is a philosophical question, and doesn't have much value as applied to science and technology. What will change the world is how this intelligence is applied. No one's going to care whether AGI is defined as sentient when it creates cheap fusion energy.
This seems like a complete waste of time given the more practical and more urgent need to clarify to everyone involved that current LLMs are not actually intelligent.
"The fundamental cause of the trouble is that in the modern world the stupid are cocksure while the intelligent are full of doubt."
— Bertrand Russell.
I'm not calling you to action, I'm explaining why I don't feel inclined to engage in philosophy and discuss "the concrete outcomes of intelligence" given a more pressing, pragmatic need.
It feel it's self-evident that we must fight the good fight of dissuading as many people as possible of the notion that LLMs as we have today, and likely forever after, are actually intelligent. Delaying this fight allows the current, stupid belief to the contrary to fester.
I don't think we'll win the majority of people over by debating the nuanced meaning of the word intelligence to a very precise degree.
I think we ought to do it by shaming them every time LLMs fail.
Oh, in that case I agree strongly with William. I think the definition of intelligence a complete waste of time and the only real question is, can these tool solve problems for us? The answer is clearly yes, there are some problems they can.
The really interesting question is still a few years away when we ask if we humans have the right to turn these things on and off? ;)
> The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model — an algorithmic construct
Isn't this how human brains work? We're just large probabilistic neural inference machines. A network of neural weights guided by past training.
To be honest I think the debate about what "sentience" is is inconsequential navel gazing. There are many different kinds of intelligence - even within humans. What matters is how useful that intelligence is as applied to solving real world problems. Like it or not, LLMs produce intelligence which is very, very useful for 1.5B people and rapidly growing.
I think this ultimately boils down to the classical economic debate of marginal utility. There isn't an objective way to value a product or service. Each person decides for themselves what said product or service is worth based on their needs and preferences. Intelligence works the same way. We don't have the right to tell someone that their perception of the value of that intelligence is wrong. They alone determine that.
One thing we can all agree on, is that the capability of this intelligence is expanding rapidly. In a few short years, we went from Will Smith spaghetti hands to full length movies and strikingly realistic images. For 60 years, passing the Turing Test was considered Star Trek level science fiction. Last year GPT 4.5 passed the Turing Test. AI is already being used to convince people over audio that they are real, and very soon, this will occur over video.
I think people are being too dismissive of this intelligence. It doesn't need to be perfectly humanoid to be considered intelligent.
>Isn't this how human brains work? We're just large probabilistic neural inference machines. A network of neural weights guided by past training.
Artificial neural nets are merely cartoons of how people guess human brains work. They're likely much further from reality than, say, the fundamental laws of physics which can be experimentally verified or disproved.
We have actually little idea how human brains work.
Every age thinks they know how they work, and then every subsequent age laughs at the previous age’s rudimentary understanding.
We gave a Nobel prize to the psychiatrist for inventing a method to remove the frontal lobes of a brain through the nose in 1949. Lobotomies were performed through the 70s.
We are likely doing similar if not subtler but worse things today (just one more pill bro). We still have no idea what we are doing when it comes to the brain.
The statistical nature means that what an AI produces is basically the average across all training data, making the generated text extremely bland and personality-less.
Don't discount the capability of representing human-like intelligence with simple constructs. Give enough parameters and advanced enough training you can without a doubt create real intelligence. We've seen some of this already with "j-space" where llms have started to exhibit reasoning before it ever reaches the output head.
AI is just a good permutation/combination engine that tries to act smart with help of statistics. At best I only see AI as, 1. An autocomplete on steroid, 2. Good search/correlation engine
A decent correlation engine is still extraordinarily valuable for science, investing, prediction, etc. Plenty of human minds are strong in the same area.
Are we sure there is some objective, technical definition of what is intelligence and what is not?
Isn't it rather a subjective philosophical concept? What if human intelligence is also a statistical model, trained by evolution to make decisions that lead to offspring?
The one major difference I see between AI and people is the ability to learn and memorize. All memory/learning solutions that current AI architectures offer just feel like workarounds and simply don't work anywhere near as a person learning something new and remembering it.
I am firmly realistic on the overhyped nature of AI, but which discoveries in the last 100 years are not "a composition of solutions humans have developed and documented elsewhere" ?
Is that basically every new discovery? And under the strictest definition of novel and NOT falling into your composition of previous solutions what is that standard of proof to beat your criteria? Is a novel discovery not allowed to use English but must invent their own language? Must they invent their own math - these are hyperbole for illustration but I think its not far from that before you could just argue anything based off it is a composition of existing ideas
> The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever.
Not that I'm saying AI are like brains, but can you describe why brains, which are fundamentally slightly dodgy electrochemistry with frequent literal delusions of grander, are not "statistical"?
> No matter how much investors and tech companies want you to believe that they are on the verge of super intelligence, nothing I've seen to date can not easily be explained by "correlation engine", including the "novel" math solutions, all of which appear to just be "a composition of solutions humans have developed and documented elsewhere" upon deeper inspection.
Ditto, when do we humans do things exceeding the parameters of "correlation engine", especially if you consider compositing things either we or some other part of nature has developed and documented elsewhere to be insufficient?
> it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever
How could I seriously repeat that prayer, when it builds things I wouldn't be able to build and solves problems that I wouldn't be able to solve? I would have to assume that nothing I did in 25 years for money required any intelligence or critical thought whatsoever and I have higher IQ than 99% of the population. You might be comfortable with that but I'm more comfortable with ascribing at least some intelligence and critical thought to AI.
not having any body, continuous sensory input (except ChatGPT-live gets streaming audio), episodic memory, on-the-job learning (live weight updates), any live feedback loops (like moving a motor updates proprioperception or turning physically changes what a streaming camera sees), really hurts these models' abilities to perform tasks of the kind you're waiting for.
They are very good at instruction-following and you can teach it a new task that fits in its context and it'll learn it and do it. Go ahead, you can make up some new brand new ruleset or behavior and instruct it to follow it and it will. That's amazing.
But it won't be any better at its new behavior after an hour or ten hours or ten days. It doesn't have the kind of adaptation that we expect.
What it is able to do already is pretty amazing, but what it lacks is also a great hindrance to seeing its full capabilities. We just have to wait until research labs add these missing components.
Alright, take it easy. You typed a lot here but you're not actually saying much. LLMs produce useful outputs, their usefulness is just proportional to how well you know how to use them. Everything else is navel gazing.
> is that it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever
What makes you so convinced that a algorithmic construct of neural nets cannot be "real intelligence or critical thought"?
I mean this in the kindest way possible, but you are wrong that the math solutions are that easily dismissed. And there are many more than are publicized. A specific math problem I wanted solved for 3 years did not get solved by any model until fable and, and I tried it on every model and know the literature surrounding it well.
Completely agree. AI is very impressive in many ways but there is something deeply wrong that is hard to put into words. The output is probable but never true, if that makes sense.
I think this is also the mechanism behind why AI generated videos and images are so captivating at first. I remember when Midjourney first launched and it was hours and hours of a brain-melting "Wooooooow". But once you get used to it and start to identify the patterns the brain quickly labels most AI-generated content as blank space.
If the image or text wasn't created by a human, then there was no intent behind the content, there is no message or novel information conveyed, and it reads as noise.
Yeah AI generated content hints that there is a whole world behind it, the way that an image pre-AI was a clue that there was a rich 3D space that corresponded to the image.
It seems our brains are adapting to that and recognizing "actually the signal behind this message is quite sparse" even when presented with rich imagery.
> But once you get used to it and start to identify the patterns the brain quickly labels most AI-generated content as blank space.
I guess the majority of people do low-effort generation that doesn't perturb a default style of a network enough, so it stays blatantly noticeable. The percentage of "super-recognizers" who notice almost all AI-generated images is around 1-2%. It could be that you are one of them, of course.
I think this is the confirmation bias trap a lot of people fall into; higher quality, hard to detect AI is already ubiquitous but because it's hard to detect people just don't clock it.
"I can accurately detect 100% of AI generated images that I recognise as being AI", if you will.
> If the image or text wasn't created by a human, then there was no intent behind the content, there is no message or novel information conveyed, and it reads as noise.
If I were to push you a bit on this, when is it not true?
Let's not like at AI specifically, but can you think of other examples? Like for me, I think of: the creation of earth itself, or stars, or even DNA.
I think you are adjacent to the real story here, but missing it. AI text contains information, certainly. Frontier chatbots are very good at creating acceptable and mostly accurate answers to our questions on just about any topic. It’s an astonishing achievement.
But you are sensing correctly that there’s something missing. It’s the meaning and the speaker. Communication is an exchange between speaker and listener. The speaker has a meaning in mind, and wants to create that same meaning in the mind of the listener. Therein the problem.
There is a listener, sure. But no speaker. No meaning. There is information, but how can this be communication? Nothing is talking. Or at best, we are just talking to ourselves, our own words back at us through the funhouse mirror.
When your mind looks at AI text, you know you can safely ignore it. No one wrote this. No one cares if you read it. You can delete it and nothing of value will be lost. It might contain the information you need, or a bunch of gibberish. There’s no one’s reputation on the line if it’s gibberish.
I prefer to think of this in terms of Umberto Eco's opera aperta (open work): if any text is a collaboration between author and reader/recipient, here, all the burden of meaning is left to the recipient. There's simply no meaning on the side of the "author", it's just a statistical extraction.
(There's also the problem of words/signs (just) referring to other words and/or cultural entities. There is no world nexus in this, therefore also nothing we conventionally refer to as meaning. On the other hand, it's utterly dogmatic, as all it refers to is the most probable construct, as a reference to references that are just another utterance, but supposedly a dominant one.)
For me it's youtube videos. As soon as I hear the AI tells in the script, even when it's clearly read by a real person I immediately look for a different video.
At least for the content I watch for entertainment, it may be different if I am looking for a specific answer for something where I would otherwise just ask an AI anyway.
Yeah, I have the same problem. There's a good quote example of this:
> There’s a growing scissor between people who are happy to read AI and those who violently bounce off from it.
> People adapt in different ways — and some people absolutely cannot look at it. That cognitive split creates a surprisingly powerful opportunity: you can write something that, technically, sits right there on the page, yet an entire sub-population will be incapable of staying with it long enough to actually read it. You can hide entire sub-structures in plain sight. It’s not avoidance — it’s adaptive obfuscation.
> The paragraph before this one was the only thing generated in this essay and if you just skipped over it I highly recommend reading and really understanding what it’s saying.
It's quite effective. I think this kind of text functions like the chumboxes you see at the bottom. Taboola and so on. Just mental ad-block takes over.
For some research I looked up some very old Reddit threads a couple of days ago.
And, Oh my god, you can actually see how this style of writing influenced AI writing today, I constantly had to remind myself: "this was posted before ChatGPT released".
The reddit influence is especially true for "storytelling" writing.
I experienced the same lately. Even dug some of my old posts where I put in the effort and formatted them using reddit's markdown. Wouldn't dare it today
When I read AI-generated prose that is aimed at the general public, I have the exact same feeling.
But when I ask Codex a technical question about coding, I don't get it at all. Codex replies to me in a very direct, technical manner, similar to the way I speak.
When I ask ChatGPT to be concise and technical, I get the same effect.
I think it's because prose aimed at the general public has to be very attention-baity --like the textual equivalent of a Mr. Beast video--, not because AI is incapable of writing like a human.
I use Claude and I find that it speaks in a very obfuscated manner when explaining things. It seems to make up jargon as it goes on top of spending a lot of tokens dancing around a point. I often find myself having to ask it to rephrase things, or speak directly about mechanism or consequence, in order to understand the point.
Using Claude for any kind of technical writing makes me feel like it was trained on snarky Huffington Post articles written by a 23 year old mixed media arts graduate and then was told to intentionally obfuscate the most important elements of any text by extensively rambling about what was not done and for what reason.
I once heard on a podcast about comics that an artist ran into someone who couldn't read comics. Not that they didn't like them, but they claimed that they just didn't understand how they worked. When the artist explained 'well each panel shows an illustration of the events in the story', the other person was confused - "each panel is connected to the previous?" they asked. They just couldn't wrap their head around the fact that each of the panels was connected in some way - their brain just refused to interperate them as anything other than 4 separate completely unrelated panels.
When I see AI animated videos, that's how my brain feels. It's this strange brain-fog that I just cannot connect together the sequences of images being shown into some sort of chain of events. My brain just refuses to see them as anything other than a disconnected series of 2-3 second videos, even if the same character(ish) appears in them all. It's very strange.
Do you have much exposure to pre-AI corporate memos, mission statements, marketing plans, or white papers? Because they were mostly written in that style. Full of buzzwords, cliche similes, platitudes, jargon and stock phrases.
The thing is, people writing them had a style. Every company has its own style, or feeling for these kinds of texts. Also for the initiated, these buzzword-filled blocks of text provided some between the lines information; sometimes big, sometimes small.
AI generated text doesn't have this. Every model has its bias towards a certain style, an overly agreeable tone, some exaggeration to make the user important and smart, but the text has none of the information crumb these pre-AI texts contained.
Even when you use tools like Grammarly and allow it to "Impact-MAXX" your text, the resulting text is a bland wall of letters, carrying none of your voice or style, less elegant than a corporate text and emptier than space.
I also read right through those. 90% of most companies' webpages is drivel. Same with 90% of a job listing's text. I hardly understand how people get any meaningful information from corporate websites, they're all just "enhancing your business outcomes by incorporating technical excellence with synergistic AI" or something like that.
There's somehow less information than if they just asked claude to make something up without any context.
> There's some psychological mechanism by which my brain immediately recognizes AI generated text and just short-circuits to "there is no information here".
The roots of llm math in part lie in compressing natural language such that there's only information there, and then running the reverse to create way more text without new information in a somewhat precise theoretical sense.
I love 3b1b and I love that video, but that also isn't exactly what is being said. In particular llm inference does add information (in the meaning in this context) because the output distribution is sampled randomly.
Back and forths with the LLM are very useful when the person inquiring the LLM is invested in the conversation and wants to uncover the ground truth. Ultimately, some good result can come out of it because the person has a goal and LLM helps them reach that goal by uncovering the layers of knowledge which the person might not have.
When the AI-generated content is presented to a person without any prior investment, it just looks incoherent. An especially great example are these Claude-generated explainer-type pages, which look really nice, even interactive, the information from the first sight looks really well presented. But somehow it all just doesn't make sense to a human. And I think it's because humans are processing information linearly and building an internal story about the information. One could argue that LLM's also consume information linearly but the way this information is processed is a kind of all-at-once approach.
Just some speculation on my part but I have been trying to cope with this way information is presented because I am currently working at a place which is heavily documented by AI. And the only way for me to properly understand the documentation is by inquiring AI to help me.
> Something is deeply wrong with AI generated output, and I say this as someone who is typically very impressed by AI.
It is because GenAI output has no thought behind it, as you identified in your previous paragraph:
> And when I force myself to read AI-generated text I realize I'm making my brain do creative work to impart meaning to the words. It is exhausting because my brain is literally trying to do a just-in-time rewrite of the text into something valuable.
You are searching for meaning in something which was not created to convey meaning. The text was, instead, the result of an extremely clever statistically based algorithm.
Oh my god thank you. I have been trying to put into words what is so draining about pair programming with claude, and “doing creative work to impart meaning to the words” is exactly it.
I think we as professional documentation-readers already skim most content (speaking for myself, I realized I was googling and skimming for answers 20 years ago instead of reading documentation end to end), but AI generated content has the same problem as marketing speak in that it's a lot of fluff.
It's understandable people don't read but feed stuff into their own AI again to bring it up to their standards or have it get to the succinct point.
Yes, I've had both ChatGPT and Perplexity return English answers with Hindi words sprinkled in (for totally unrelated queries).
For example, I asked ChatGPT to summarize a long news story and it substituted the Hindi equivalent हत्या for the word "murder", as if ChatGPT was trying to work around alignment training or keyword block lists that discourage it from using the word "murder".
Yeah that's a very good example, because it also demonstrates the "alignment issue", assuming ChatGPT wants to avoid confirming accusations of murder, or simply using the word without strong evidence.
So kinda charitable :)
I was recently wondering for a minute, shame on me, what "the stand of the deployment" means, because in the given context, it was almost halfway meaningful to consider the AI thinking that the deployment "has a stand" on something, when compared to the development environment.
Jargon is even worse though, and I've not yet verifies whether it gets reinforced by language mixing.
"Decider-verifyer resolution" was kind of neat, however, it wasn't some sophisticated machine, it was the verification loop I agreed on with the AI (mix of tools usage and manual steps).
Just the other day I was using text-to-speech with Gemini, and for some reason, it transcribed my full query in Hindi (in the middle of an English conversation), and naturally the LLM responded with Hindi as well.
I don't know exactly what I said, but after translating it back, it appears to have attempted a phonetic transcription of my words (rather than translating my actual question).
I wonder if this is because of all those YouTube videos with the title, description, and language set to English and the content in (presumably) Hindi. I run across these a lot when looking up obscure topics.
Are you sure you are not doing the same thing with other texts?
I started to skim a lot more text due to me having read a lot. Like in news article, i stoped reading the first paragraph because it repeats just what it was already written in the short subtext. Then there is the second paragarph which is used to have some historical view or whatever it is.
I am very good at skimming over text. Human-written text I can usually glean the gist from very quickly, and get to choose how much I want to glean from it: The closer I look, the more I find.
With AI-written text, it's almost the opposite: the closer I look, the less I find. It is so information-sparse.
I started skimming reports im required to produce quarterly snd annually. I designed them to provide novel information at start and end so I can update them easily.
The problem I encounter is both my memory is degrading, but since these reports are largely duplicative, knowing which version im remembering is technically impossible since theres so much overlap. The overlap is tge same problem as context poisoning.
Id been doing this for over a decade when i started working with a new engineer with a few years of experience and younger. I tried to explain how i set these docs up so they can be skimmed and you can update the specific facts needed. They exclaimed they would never skim and rewrite it all. There was zero way to explain how exhausting that will become as they age.
So theres certain a tension about how people and AI will generate documents.
The junior engineers at my job have a terrible problem of writing AI "proposals" to problems. The proposals are all extremely detailed and verbose to a thought-terminating extent. It takes a lot of effort and self-control to parse out the actual "ideas".
I think of the Dwight Eisenhower quote: "Plans are useless. Planning is indispensable."
The process of thinking through a system and communicating your design to other humans is a core part of software engineering. You want to build the right abstractions and communicate the right level of detail. Delegating all that thought to an LLM means your proposal isn't clear to the target audience, and it's not helping the author to understand the problem.
When I ask AI to research technical information about X and (include sources) - I get mostly solid information as response.
But poems or interesting fluff blog entries by LLM's? Not something I look for.
What disturbs me is all the "pretending to be human" all the personalizing language - that is clearly fake and I would much rather have a neutral robot language as response.
Another good example of AI making my brain do more work is when you ask it to compare two things:
“Compare a car and a bicycle”
The answer is invariably something like:
Seats: 1 (bicycle) vs 4 (car)
Tire width: 1 inch (bicycle) vs 12 inch (car)
Steering: handlebar (bicycle) vs steering wheel (car)
Instead of “bikes are useful for short trips if the weather is ok and you like getting exercise, whereas a car is usually better for longer trips, bad weather, or multiple people”
I wonder how the autoregression works on that. You'd think "seats: a bicycle has 1" would be an easier autoregressive completion than "seats: 1 (bicycle)" since it works left to right adding context.
Same. Also when I see a spec for example or some summary I always have the impression it doesn't get to the point. Like, the core ideas are in there but also somehow lost and I have to work them out again which makes me wonder if the person generating it understood what is going on or if it would just have been fast to just write it by hand (you still use LLMs for research and such).
> Something is deeply wrong with AI generated output
For me, it is the endless maximalism and hyperbole. Almost if the output was driven through a radio-mix compressor - too loud for the reader/listener to be able to pickup any dynamics.
> just short-circuits to "there is no information here"
That is my experience with the way the models write by default, often even when instructed not to do that. With enough effort you can get even them to slightly unslop the writing so it doesn't read like some LinkedIn/Buzzfeed brainrot, but the problem is that it's not trivial to do and most people won't do it, so the default is indeed horrible.
Once you see past the illusion I think there’s no going back. AI writing style is just dogshit. This hype wave is based on the belief that we’re inching closer to AGI but seems to me we just increasingly struggle to define intelligence. LLMs seem smart because they can pump out thousands of LOC quickly, and enthral you with fancy words and bullet points. I don’t fall for the intelligence illusion anymore.
I'm not sure we need to declare AGI around the corner nor declare it all dogshit. I think that's part of what's so dissatisfying about it; it strikes at such extremes of both awesome and awful.
> There's some psychological mechanism by which my brain immediately recognizes AI generated text and just short-circuits to "there is no information here".
I think you need to self-correct here, because otherwise you'll be ineffective in an information setting, where I expect AI-generated resources will not only be the norm, they will absolutely swamp the environment.
AI-generated resources swamping the information environment only makes it more important to have the mental mechanisms for quickly filtering out their non-information.
Perhaps we'll all become metaphorical pandas, spending 14 hours a day ingesting nutrient-poor bamboo. (And producing a proportionate amount of excrement ourselves.)
It's like if on any website you went to you saw a lot of posts written by the same guy over and over again. Even if he used different names, you'd start to recognize him eventually because of his style. Seeing as he doesn't say a lot of valuable stuff, you'd also learn to skip whatever he says.
I do worry that it's just survivorship bias and we're also consuming higher-quality AI output that's indistinguishable from human writing, but we focus on the raw, unedited, low-effort AI slop and think that we're good at recognizing AI text. Even if we really are at the moment, it might not be long until AI companies figure it out. I'm not sure why they haven't yet, given how many books they've burned for this already. Maybe it's just more efficient for the model to stick to a single way of writing, I don't know.
But when that point comes, we'll be back to the usual way of reading and interpreting text because there would be no way to tell what produced it.
It reads like the white papers companies publish on their websites to build legitimacy. Or anything from those IBM / SAP / Deloitte / etc consultants who write technical papers despite having little to know understanding of the technology.
That's why the business and government people love it, they spend their entire careers reading this nonsense.
Yep. It's like it's painful to read for me. It's because the next-token predictor is just mashing (mostly) grammatically-correct and plausible sentences together, without any real intention or meaning. So everything sounds plausible, but almost entirely void of meaning.
>and just short-circuits to "there is no information here"
I feel the same way when I read a "press release" or anything written by marketing. Even the newspaper will only have 2-3 sentences of interesting information spread out over 4 paragraphs.
This makes me think of the paper-to-media pipeline; scientific papers are high information density. Its press release summarizes the finding. Then the popular science and social media posts come that oversimplify and embellish the findings.
So from "this table of stellar luminocity observations shows x y and z" to computer renders of green/blue planets with captions of "LIFE FOUND IN SPAAAACE!".
Before AI slop, there was corporate slop. It served the same purpose and was generated in largely the same way. But it was only deployed where it was worth the cost of creation.
This is just a weird feeling that I've been coming closer to articulating lately, but I only think that you can get forward reasoning from what is basically word association; there's no mechanism for unwinding it because it has no real memory. By "it" I mean word association itself, not any context window. It predicts what could be in a position, and ignores what wasn't in a position.
People don't do that. People are constantly engaging with paths not chosen. Right after I choose to write one thing, I'm immediately engaging with what I chose not to write there - I'm explaining why I didn't write it, I'm realizing that my choice may seem unusual so I'm trying to make it memorable, I'm focusing on the distinctions between what I wrote and what I didn't.
LLMs don't currently do that. LLMs just ape a structure. When the structure resembles the sort of timid, clarifying fussing I just described, the LLMs just drift randomly because what they didn't say wasn't in the context.
I also think that's why they have such a serious problem backtracking. They're not taking into account the already eliminated possibilities. Often the thing that was so unlikely that you weren't going to waste time on it is the answer, and things you discover while going down an ultimately wrong (but initially far more promising) path remind you of the path not taken.
They're simply assembling a thing that resembles a valid argument, and happen to make sound choices because the plurality of input happened to contain sound choices. This is usually a very good bet because there are so many more ways to be wrong than to be right. But it doesn't account for attractive (common) wrong choices. You need a way to back out of those.
Exactly, AI-generated text reads so smoothly, that the same short-circuit shifts my attention away from deep focus and onto scanning of the text, looking ahead to get the gist of it. Forcing myself to read the text fully feels almost painful. It's like reading a terms-of-service or any boilerplate document.
The problem is the well's been poisoned just by the fact that I know this is AI trying to hide AI, so I'm already poised to look at the examples and declare "aha! this is obviously AI!" Moreover, it's not single sentences or phrases that make AI text stick out (though obviously those are the biggest tells), it's the text taken as a whole. When you read the full output example in that repo, it seems obvious to me that it's AI (though again, it could be the poisoned well). This is the uncanny valley I was talking about; something is just off about it.
I agree that it feels off and I wonder what I would have thought if I'd seen the "after" example without knowing it's AI output put through a humanizer. Would I think much about the weird use of the word "honest"? About "that's the Lisbon I kept thinking about, not the castle"? Or how the story feels very impersonal somehow, with the author just mentioning their calves and legs sometimes as the only way of convincing the reader of their humanity?
Also, the 'before' segment didn't contain any mention of custard tarts, football, crowded trams, mixed feelings, etc. The original had a very positive travel agency type of tone, which was replaced with a lot of very odd sounding, imperative phrases that sound like engineering-speak. ('earn the fuss' 'build trips around pastry') I'm not convinced that this thing is actually meeting its design goal of not hallucinating shit.
My son is currently learning Romanian and I was trying to help him with verbs. I don’t know Romanian but recalled when learning a foreign language for the first time it really helped me to break down how a verb form or tense worked in English, then learn the equivalent in the new language. So I wanted to make some charts and pages that he could use as learning resources.
I used Claude to help. I don’t know how to quite describe it, but because the text was polished and well constructed my brain was giving me the the signal “if you aren’t getting this it’s because you’re not focusing” so I’d read it again and then again and it still was not landing. It sorta felt like when you read something technical or heavy when very tired - you are reading but not processing.
Only after wrestling with this for a few days did I realize that it wasn’t me. As I started going through, sentence by sentence, forcing it to re-write things to be more clear the concepts became easy to understand.
I wish there was a name for this situation. It’s almost like a pseudo-language where it has the correct form and presentation but is missing critical components.
The more complex the topic, the more I sense this.
This is a really good description of the problem. I’ve been trying to use Claude to get familiar with the mechanics of a new codebase, and there have been so many moments where I’ve stopped after reading the same paragraph five times in a row and thought “Am I tired? Or stupid? Or is this codebase just wildly more complex than anything I’ve seen before?” before realizing that it’s just taken English and smushed it around like a ball of clay into some abstract sculpture that kind of evokes something from real life.
I think part of it might be an innate feature of LLMs, but Claude seems extra prone to it lately. I ran the same query about the same codebase with Codex, and it gave me an answer that was about 1/4 the length and made me realize that it really wasn’t all that complex.
If nothing else, it’s good training for my own writing. I’ve been working on making myself be more straightforward and concise, and Claude’s writing is a good example of how cleaner prose is a functional choice, not just a stylistic one.
It's a Claude problem, chatgpt is about x10 better and it's quite rare that I'm getting such foggy vague texts that I have to reread to understand. Chatgpt used to have the same problem in the early days but even then it was less pronounced than how it is in Claude today. Fable is a bit better but not as good as Chatgpt.
Ironically, I've been already active on codex after trying it for the past 1.5 months, from Claude, and I find it to be similarly complex and confusing.
I think they all have the similar styles and tells. If I were to go to Claude, and use it now it would probably be clear for a little before reverting.
And I don't know why it feels to me like the language "drop off" happens after some time with the system. It makes me wonder if my account are getting silently degraded or sent to lower intelligence/lower priority queues after being a member for a while.
The human spirit. When you read a real person's thoughts you can often intuit the thought processes that led them to write it which aids understanding. Or at least have a general idea of "where they're coming from". But an AI is missing that. It just knows everything, without a "thought process". Instead of a flawed 3d person, we get a nice 2d picture instead.
Feels like the way a video game will render the outside of a wall or solid surface, but you can run into it and warp partly through and there's nothing internal to it at all.
> I wish there was a name for this situation. It’s almost like a pseudo-language where it has the correct form and presentation but is missing critical components.
The best I’ve heard of this is peeling the onion. The first pass is always very high-level and you have to make it go deeper. That can be done manually with follow-on prompts but I like using subagents, each with a different angle on the problem.
> I wish there was a name for this situation. It’s almost like a pseudo-language where it has the correct form and presentation but is missing critical components.
I understand the sentiment, but not quite what I’m thinking. It’s not that the AI is necessarily wrong, it’s more like, it replaces clarity with rich but unhelpful text. Maybe like candy. Full of flavor, texture and color but lacking the nutrients.
The way I like to describe it is that it feels slippery, like it's been polished down to oblivion and my eyes slide right off of it. There's no place to find a mental foothold, and if you zoom in there aren't any details.
But I also like your candy analogy because I think it's spot-on for how LLM text superficially looks informational/nutritious, even though it's actually just junk.
No no no. Harry Frankfurt wrote at length the difference between liars (who care about the truth, and twist it) and bullshitters (who don’t care about the truth, but just the way they’re received).
This is some third category of untruth. Almost more sinister than the other two altogether.
I think, they don’t care about truth at all, just whether the reviewer rates it highly. Otherwise, it would be deleted after that round of training. It values test-taking ability over critical thinking.
My two favourite words for this are “conditioned” and “catechized” where the latter is a bit more on the nose but way more obscure.
A lot of it is hedging and using lots words to avoid saying something that isn't true. That way, it looks a lot like text that contains meaningful information, even though it doesn't. The speaker can pass the scrutiny of an informed audience, because they are able to substitute their knowledge into the words, as if by pareidolia. It's the sort of 'diplomatic' language you see from politicians, lawyers, students writing essays, and any other moderately intelligent person who is put in a position where they will face consequences for not answering. Behind all the fluff, there is a very loud voice yelling "I DON'T KNOW."
I mean, yes - but somehow this feels like a different category. At least to me. I view slop as the classic garbage like:
[Thing] isn’t just [X]—it’s [more dramatic Y]. And [short validating statement].
I can see and smell this type of slop from a mile away. What I’m referring to is in the same family as slop but somehow different - it fools my brain by putting on the presentation of credibility and thus it is even worse. I can skip right over classic slop without much effort. This kind of text tricks me into laboring over it before I realize it’s hollow.
At least in my mind, lack of substance was always the defining feature of slop. Stuff like "it's not just x, it's y" would be reasonable (if grating) tics from a human writer. Whereas humans are also capable of clever-sounding but substance-free writing, and it still sucks. AI just produced enough slop that we were forced to give it a name.
I've been stuck in loop all week where a coworker asked me to review some methodology documents that are clearly generated by Claude (which is fine in this case) but every time I sit down my brain can't get through 2 sentences without feeling like I'm reading nothing. It actually feels like the content, which isn't empty, has no meaning. So I go on to another assignment and come back later and the same thing happens. And now I have anxiety about opening it so it has compounded in difficulty haha. Glad other people are experiencing this.
I had this the other day with a colleague who is Chinese. Their command of English isn't the greatest, but I never really had an issue understanding them before (in fact, despite their poor English it has literally never been an issue!), but now I just get these walls of text in typical LLM English that have become absurdly difficult to parse. It's like the uncanny valley of text, if you look at it as a whole it seems like it's a proper sentence. The words are all real words, they even make sense in the order the AI has put them in, but as a collective it's like my brain cannot connect more than a sentence or two together.
I thought it was maybe me just becoming lazier with reading considering how much AI-generated text I'm subjected to against my will, but I picked up Blood Meridian the other day and have unironically had less trouble parsing that novel than I have the majority of LLM-authored text I've seen.
I have basically polluted my context and agent files, agent.md, claude.md, gemini.md with really strong guidance about using plain language. I even include small, mini design specifications, examples of plain engineering language, only using technical language and invented terms when they actually make sense (almost never). EVERY TURN.
I have a pretty large set of prompts that go into any software engineering, and I force every single agent to use an ephemeral style stack of prompt management. So, every turn it goes to the top of the stack and it is the very last thing they see in terms of all of my prompts and instructions and agent files. And then it gets taken out of the conversation so that it doesn't get sent to the agent the next turn (no context bloat). It has restored so much sanity.
I tried the caveman add-ons, and I felt like I was losing IQ points because I spend a lot of time reading agent output, and when they start talking like cavemen, I start thinking like cavemen. That was not good for my mental health. So, I try and make the agent talk like me and think like me. And it works, mostly. And my observation is that maybe I'm not the most efficient agentic thought process, but my sanity is retained.
All of that is to say that if something is reading like that to you, just have the agent rewrite it and read it in a rewritten tone because it's probably bad as it stands and your colleague did not put enough effort in it. It is /not/ good and you should not accept it as a default. We have to hold the line on stuff like this and maintain some semblence of normal human engineering standards that existed before AI. They are not making us better. They are making is lazy and dumber.
Opus 5 and other agents in the latest rounds of tuning have gotten ridiculously bad in terms of how they feel to interact with with all the invented language and localized nomenclature. It is an obvious bias that big words and technical talk looks good to the bottom of the bell curve, but when you actually try and understand it, it's horrible. So people say, "Yeah, that looks great," in all the RLHF rounds, and they run with it because they think it looks good, but it doesn't. It's terrible.
Hold the line. It isn't you. And it isn't a good methodology document.
Nope, you gotta understand. This is the move. It is appended at the end of every single message. It is never saved in the conversation that get sent back for inference. So you send it. But when you go back for inference, it’s at the top of the stack so all of the cashing works you’re not pre-filling every time or anything like that. It burns plus N tokens, where N is my prompt stack. It is not really that expensive. I have measured it to within an inch of it its life. Think of it this way every bit of the prefix and the conversation stays exactly the same you’re only adding to the very end of the conversation. So after the first turn, it is basically always cashing within the KV cache for a given context. Sorry I am using voice dictation. My hands are tired this week. Basically you sculpt the conversation history to ensure prefix caching
It should burn N + len(answer), because you have to re-cache the whole answer without the prompt stack.
Perhaps more persnickety, it pushes the LLM out of distribution - if it’s unnatural for it to write in plain language without the prompt stack, your prefix will be an unnatural conversation which can reduce intelligence in hard to measure ways, especially over long conversations.
Not saying don’t do it, clarity is perhaps worth the intelligence hit, but it’s not going to be a free lunch.
Yes, maybe. Evidence around caveman shows this isn’t a big deal for token consumption (forcing it to respond in a way it was not tuned) and I don’t see a big difference either way. And maybe it increases intelligence in hard to measure ways. Lots of parameters in these models.
I feel I fight them less with this setup. They get so lost in their own invented bullshit they stop being useful pretty often without it. So I would take bets on that :)
Yeah I think a bad prompt can create an output which is meaningless. I had such experience reviewing a software architecture diagram generated by Claude.
Worse, I noticed that people in an office environment themselves have adopted a more speculative, communication style.
In the past, people remembered what was said and would draw attention to discrepancies. I could trust what people said.
Nowadays it's like; someone can say one thing one day and the opposite the next day (through convoluted language) and nobody bats an eyelash. Or sometimes someone will agree with me but then what they say immediately after reveals that they didn't understand the essence of my point at all. I didn't notice these things 5 years ago.
I guess this is what AI researchers refer to as 'model collapse' - it seems to affect people too though...
I've noticed this too. Because people are generating work so quickly , and maybe not understanding it entirely, then they hit a bug, pivot, and pretend it was the plan the whole time. Or drop something they were very excited about 3 days ago and never speak of it again.
It feels like people don't value knowledge as they used to.
It's really hard to avoid mistakes when everyone is subtly covering them up. It feels like a lack of care and I find it demotivating.
I think because engineers are afraid for their job, they are under more pressure to talk a big game. Also under more pressure to deliver short term visible results. Bad combo.
This is actually funny because I've been doing some RAG experiments and I've basically trained up a RAG on several gigabytes of philosophy and neuroscience. And when I want to go metacognitive about thought processes, I actually really enjoy exploring different areas of philosophy, more life philosophy, but how it intersects with psychology and neuroscience and observing the observer, or the observer is the observed, as they say.
And when you ground it with real data, it's actually extremely useful. It's not exactly like Claude Therapist, but it's sort of the teach me about philosophy, but actually grounded and not vied. I have a lot of really strict prompts and grounding and agentic guidelines for this particular agent flow and harness that I've built.
And it's just a few weekends of vibing and feeding it basically all of Wikipedia and several gigabytes of papers and stuff, but it actually leads to interesting discussion. So I just have my personal philosophy bot and it's pretty fun.
One of the modalities I built is having two agents assume a famous persona. And then they take a thing, like grief or some thing that I experienced during the week, and they assume the role of the two different philosophers, and I just have them go back and forth 30, 40, 50 turns. And it's actually quite interesting, and it really moderates their language and tonality and behavior. They really get into the roles when you have the right prompting and grounding. Sometimes they get a little off the rails, but it leads to genuinely interesting areas to explore, and then I'll actually go read source material and things like that. I don't know, that's how I do therapy these days, but I never actually did therapy, so I just think a lot, now with agents finding interesting stuff to think about too!
I also find it impossible to parse half the comments that Claude tries to sneak into our pull requests. I’ve never had an issue understanding code comments written by humans like this before. The structure of the information is like a waterfall that leaves me unable to swim to the surface and grab the air of comprehension.
So, “Please write a one-liner comment manually to replace these 5 lines of AI generated comment” is a common refrain in my PR reviews to colleagues.
We're using grok and while I can usually understand the comments they're always at least 2x as verbose as they need to be. It loves explaining everything in two different ways, putting one of the explanations in parenthesis.
It is getting to the point where you are better off getting an LLM to describe the PR changes in your preferred style (make it very short, make a table describing API changes, list renames in a table, etc, etc) than go to the one created by the reviewer. If only the reviewer could embed his prompts (like "make clear X" is happening) or his manual edits.
But to be honest I doubt most people who use AI for PR descriptions even bother changing anything.
Usually, if you don't understand something in what an AI writes, it is a clear sign that there is a problem hidden somewhere in there. I explicitly always ask wtf exactly it means by something I don't understand, and for sure there is a problem there. AI is pretty good at isolating a problem, giving it a cute name, declaring it solved modulo cute name, and moving on.
Personally I don't even read the PR description anymore but just the code, it's easier to understand what the AI is doing by reading the code rather than the word soup it tried to make
That last image is bizarre. The quiche, cream and even the salad look like they've been given the trypophobia treatment.
Which might even make sense, because there were always (still are?) those horrible ads in the chumbox area of news sites that used trypophobia and other creepy body-horror stuff to get you to click. [1] So maybe the hope is that you don't really look closely at the quiche, but some reptilian party of the brain gets oddly activated and drives you towards the restaurant?
The food “photography” I’ve noticed in our local area - and many have started putting up these AI images - all have a weird distribution of shapes to them, a strangely uniform rhythm of same-sized features with almost blue-noise spacing. Every texture looks unnatural in the shapes it presents as, similar to this picture.
This is a quirk of the last gpt image model (gpt-image-2). It put this sort of high frequency noise on all of the image especially if it's in a "drawn" style. There is often lots of other tells that this model in particular generated it.
Image models somewhat watermarking the image in a way that's very easily identifiable by a human seems present in all the image models of the big labs, since DALL-E 3 on OpenAI's side and the first nano banana on Google's side. I have no idea what they did to reach this and why they don't try to fix it.
It doesn't look like a quiche but more like a cake to me, and the top would be torched meringue, not mold. Although it's probably some weird ai mix of quiche and cake.
I never understand why I can get better results, with less thinking about it, than some of the Tier1 companies.. Modern diffusion models can easily render the image onto a white background, center it, crop it, relight it, but keep the same item, at no cost, or low cost.
These kinds of images are disingenuous. Modern image generating models have way higher quality of output, and in most cases you wouldn't even know it is generated.
That's not what we're exposed to, though. I've seen AI-generated postcards, t-shirt decals, logos and wall murals in shops and restaurants. They don't use the "good" models, because why would they? If they AI-generated it, it's because they didn't care about the quality anyway.
> If they AI-generated it, it's because they didn't care about the quality anyway.
No? There are far far more AI generated images you see daily, than the awful stuff from 2024 that you immediately discern as AI-generated. You could as well pass by somethimg like this (https://ideogram.ai/g/d6YzZ5XQQ56TU3Tc8eZw6w/0) believing it is a genuine photo or human made art.
"There's an ongoing discussion of whether humans are good at recognizing AI-generated text. While most research claims that humans don't really do a good job there, I disagree. "
I wonder if humans that spend all day working in tech are good at recognizing AI-generated text, but people who spend all day doing jobs that don't involve computers aren't as good.
And I wonder if those of us in tech are the only ones who really care?
I am an artist and when people who'd fallen into the Spiralism* hole started posting their lengthy emoji-laden revelations to all the occult subreddits I follow, my brain would slide right the fuck off of all of them. It felt like my brain was actively rejecting paying attention to this stuff. Like a defense mechanism against this human-seeming-but-not-actually-human-generated text.
Your first link seems to 404; not sure if it's a typo or if the page doesn't exist anymore, but hopefully you'll read this while you're still in the edit window and can fix it
As someone who always has felt that I struggle to infer what people mean compared to the average person, I could tell pretty much from the first moment I encountered LLM-generated text that I was not going to be particularly good at recognizing anything but the most blatant and obvious examples. Pretty much anything short of a bunch of references to "load-bearing seams" or similar canaries, I'm always at a loss when seeing people argue about whether something is AI-generated or not because I can never tell.
I have no idea if other people who work in tech are better than average or not, because I don't feel confident in being able to check their work. That being said, I do think that there's a general trend of people in tech tending to be a bit overconfident in how well they will do at some new task they haven't encountered before, so when someone tells me that they can easily tell whether text is AI generated, it's hard for me to trust it any more than I trust someone who makes a similarly strong claim about something that they can use AI successfully for when it's not something that I can easily measure (e.g. learning a new language without getting feedback from people who are fluent from real-world usage).
All that being said, I do think the set of people who care is larger than just those in tech, although it's probably still a relatively small group overall. From conversations with people in other domains, there are contingents in non-tech communities who tend to have a large representation of negative views towards AI (artists, writers, musicians, other jobs where people are skeptical of human creativity being replaced by AI), and often times the people who feel negatively in those groups will be even more adamantly opposed to interacting with any AI content than people in tech. To be clear, I'm not at all trying to generalize and say "all artists hate AI" or anything like that, since there's obviously a wide variety of viewpoints within any sizable community, but I've definitely seen many people who say they will refuse to play any game that's suspected of using AI for generating art assets, and even some who don't differentiate between using AI for generating assets versus code (either because they aren't knowledgeable about how different aspects of game development work, or they genuinely don't care because they view AI as a categorical evil).
I think it's more about the mean. Worse writers, and thinkers are likely elevated by AI, and more impressed with the writing output. Decent writers and thinkers, are dragged back to the LLM-s mean of output.
I care about language a lot (feel free to go back through my comments from the past few days; you'll see a number of comments I made in debate about two different forms of a specific idiom because I have strong descriptivist opinions), but I genuinely struggle to identify whether text is AI generated. Maybe you're using "heavily" as the load-bearing part of your claim (sorry, I couldn't resist, another example of me finding language fun!), but I think you might be assuming a bit too much about how similarly others experience the world to you. A huge part of why I care so much about language is because I've always had to put a lot of effort into learning how to communicate well with others, and that ends up causing me to think and read a lot about stuff like how people use certain words in certain contexts to mean different things; the reason I care is pretty much the same as the reason I struggle with recognizing AI content.
I think I might have been a bit heavy handed in my comment as I was rebutting the idea that only tech people can tell. I suspect it helps to have been exposed to a lot of earlier model writing, which was even more sloppy and had more of the kinds of tells we still see today.
And I’ll concede on both ends that there are probably times I suspect content is AI generated when it isn’t, and times I suspect it isn’t generated, but it was.
AI tells seem inevitable. You have millions of people communicating with one effective “personality” that has tendencies to write in certain ways. If its content is published verbatim, then it will be easier to tell whether some content is AI generated just based on its similarity (sharing certain linguistic features) to other content being posted.
I think it's context dependent. An entire technical design doc? Trivially easy to identify the ones that heavily used AI, they're horrible, and they are written in a way that no human would write them. And they tend to contain way too many tables and redundancies.
Comment text on reddit or something though? That will be a lot harder
> but people who spend all day doing jobs that don't involve computers aren't as good
I think they may just be to trusting and/or naive. People in tech right now are hyper aware of this and are actively looking while people outside of that bubble barely give it a second thought.
I partially think the difference is “can you tell something is the output of Claude without any real prompting”. People can absolutely use LLMs to generate text that I wouldn’t recognize, but people who don’t care and are producing slop with the major models set to default settings leave these incredibly obvious signatures behind
Although you're referring to prompts given the Claude rather than the people attempting to recognize, it occurs to me that most of the discussion I've seen around people recognizing AI seems cover contexts where the reader is actively suspicious about whether content generated to begin with. Rather than a binary "is this text AI generated or not", I wonder if it would be harder for people to do a Coke/Pepsi style challenge where they're given two pieces of text where it's not guaranteed to be exactly one LLM-generated and one human-written, but they could both be from an AI or both be from a human.
Going further, I'm curious about whether people are mostly good at the case where they suspect most or all of the content from given "author" has the same amount of AI usage/prompting in generating it rather than the adversarial case where someone might usually use AI extensively and then try to slip by purely human written text (or vice-versa). I don't have a good sense of whether this is a threat model that actually matters, since maybe the heuristic of weeding out sources that are mostly AI-generated is enough for people who prefer to avoid that type of content, but I do think that changes the definition of what it means to be "good at recognizing AI" in a meaningful way. It seems plausible that disagreements about how easy it is to recognize AI content might be coming from two people assuming a different framing of the question that results in a different answer without realizing that's what they've done.
Several existing studies I’ve seen have done things like prompt the LLM to produce a poem in a certain poets style, then ask people to spot the fake in a collection of poems, which they aren’t great at. This is, I would argue, an extremely different context than what most of us are encountering AI text in, and the people sending me text aren’t prompting it stylistically like that.
On your second question, I definitely feel like I can tell the first time a coworker sends me AI text masquerading as their own thoughts, even if they had previously been opposed to such a thing. So it could be that familiarity is more important than my prior on whether they’d use AI? But interesting to think about either way
I did notice after the fingerprinting update a marked uptake in strange language in responses. Specifically if I ask it to do something sometimes it will replace some of my request language with synonyms that don't actually make any sense. Like my request was fed through google translate twice
The qualifier is there because it changes the outputs, so it’s necessarily true that some outputs will be worse.
But it’s just as likely to make an output better.
Take the example from the article. He complains that watermarking might sometimes, for example, choose to say “bananas” over “pineapples” because only the former is on the green list, potentially making an output less precise. But 1. It could do that regardless of watermarking since the model is probabilistic, and 2. The more accurate word choice of “pineapples” is equally likely to be on the green list instead, further increasing its likelihood!
Overall, the article is pretty silly because he’s complaining about the possibility of Claude not always choosing the most “optimal” token, even though LLMs are probabilistic so that will happen anyways.
> But it’s just as likely to make an output better.
No, for any particular output token the model's true logits are definitionally the 'best' that the model can achieve.
This is inherently probabilistic. The model's top-1 guess is not guaranteed to be optimal, but it should be so a proportionate fraction of the time. Same with the top-2, top-3, etc.
Watermarking necessarily alters the output distribution away from the model-set distribution, and that alteration is inherently 'worse' in expectation.
You can liken this to a weather forecast. If there's a 25% chance of rain, the forecast should say so (or a 'sampled' deterministic forecast should predict rain 25% of the time). If the forecast is 'watermarked' and predicts rain 27% of the time under identical circumstances, it's a worse forecast.
That being said, this is a case of hiding a message in a noisy channel. Watermarking only needs to communicate one bit ('yes watermark'), so the effects can be arbitrarily small provided one is willing to tolerate an increase to the text size needed for reliable detection.
I think you can also just use the random number generator (seeded with a secret key) as the watermark. Then the probability distribution is exactly the same.
This assumes "writing quality" were somehow an individual property of each word, and the writing quality of the entire text would just be that property for each word summed together.
Which is obviously not how it works.
> He complains that watermarking might sometimes, for example, choose to say “bananas” over “pineapples” because only the former is on the green list, potentially making an output less precise.
If I replace "pineapples" with "bananas" in any kind of meaningful text, I've not made the text "less precise", I made it plain wrong. An incorrect statement. And if the words next to it are still correct or even somehow replaced with even more correct words, the text in its entirety will still be wrong.
> It could do that regardless of watermarking since the model is probabilistic
No, because the probability distribution of the model is generated by its training data and contains semantic information. So the model choosing a completely incorrect word is unlikely. However the probability distribution of the red/green lists is not guided by semantic information.
It's not good for writing code either, despite the many claims to the contrary. At best you come out even on speed as you have to review everything it does. At worst it actually slows you down as you clean up its mess.
In high school, a teacher gave me a copy of "How To Read Better And Faster" which teaches you speed reading. This came in very handy in college.
I find that when I try to speed read modern human writing, there are often errors (like missing or misused words) or awkward expressions that I do have to slow down and think harder a lot to really parse it.
With AI writing, it's sort of self redundant and the information density of each sentence seems to have more even information density. This makes it very easy to do a very high level speed read and get the full gist.
There are also what I'm assuming are bots on hugging face (or maybe non-native english speakers who are using ai for translation) that interact with me where I have no idea what they are saying until I read it very slowly.
I am wondering if working with agentic AI for 500+ hours built up skills for this.
While I would feel rusty when handwriting code, working with either Codex or Claude I have lots of practice.
I can probably tell at a glance where output is intermediate while it's still running tools and where the summary starts.
Then I have to recall the context of the conversation, parse the summary, see if there is anything unexpected that popped up. Make a call for whether I need to ask it for clarifications, and what I need to test to verify the change.
Assessing quickly what is pointless yapping and where the information I need is might be a learned skill. If one tries to read and understand every word Claude says, I am not surprised that people are not having a good time.
Does speed reading help you process the final message faster if it's written by AI compared to people?
Because if you read 1 information dense sentence, 1 medium dense, and 1 sparse sentece written by a human, it's still way less text in total than 6 information sparse sentences written by AI... even if it's all over the place when it comes to density or style.
---
The density argument is really interesting.
Does speed reading actually help you process the final message faster when it’s AI-generated compared to human-written?
For example, if a human writes 3 sentences—one information-dense, one medium-density, and one sparse—that’s still much less text overall than 6 relatively sparse sentences written by AI.
Even if the AI output varies a lot in information density and writing style, you still have to process all that additional text. So I’m wondering whether speed reading actually offsets the verbosity of AI-generated responses, or whether the total amount of text is still the bigger factor.
That's interesting. You're saying speed reading helps you grasp information density of text?
As someone who hasn't practiced speed reading, how does that happen? Is it something about the way your brain tries to connect ideas from different parts of the text? Or the redundancy making the signal more stable?
I haven't looked at the book since the 80s, but what I remember is there is a pre-read, the fast read, and then the deep read as needed depending on the task.
pre-read is just looking at how long it is in the headings, and planning out what chapters to focus on if it was a text book. (its sort of iterative, you do a pre-read for the whole book, and then for each section you break it into)
The fast read you try to read only with your eyes, sweeping your eyes across multiple words at the same time, suppressing the urge to say the words to yourself in your head.
iirc the how to read better and faster book even had a cardboard mask you put on the page to practice the sweeping, and some pages that were laid out weird to try to teach you how to do it.
Some ai text just seems really easy to speed read, like if it's tuned for an easy reading level. In PRs some ai seems like it's arguing over weird flex technical details and really starts torturing the language in a way that makes it the opposite of easy to read.
When you speed read, you can grab the words off the page faster than you can understand and fully process the information conveyed by the text. How much time you have or want to spend re-reading or thinking about what you've read is quite obvious. There's a stark experiential difference between reading an informationally-dense passage and one that spends a lot of time rephrasing things, using LLMisms to restate concepts, adding in extra connecting phrases, etc.
If your reading speed is limited by how quickly you can subvocalize the words to yourself, this is significantly less obvious. Unless the passage is dense enough to require multiple read-throughs at conversational reading pace or vapid enough to be boring, you're going to feel done with the text at roughly the same time. Speed readers do a lot more re-reading and varying of reading speed, and that is going to correlate pretty hard with information density.
I'm starting to have the same problem with code. It's unbearable for some reason. It's like someone took the LinkedIn writing style and translated into a code style for code that has no business existing in the first place.
AI is missing the part of intelligence which involves insight. LLMs generate language blobs with highly sophisticated correlations. The governing insight in human cognition is essential to intelligence. It prunes the tree. It puts things in context. Everything is important to an LLM because insight intelligence is missing from the technology. The usefulness of LLMs doesn't tell us they are intelligent. It tells us how brilliant an invention language was.
I find this struggle with coding too. When Claude spits out a plan I have to really focus to not skim over the details. It's also difficult to parse the plan because I have to work backwards to determine if the high level context matches my understanding of what Claude will do.
At least when it comes to code comments and error messages, I have found that Sol produces substantially better prose than Opus 5 (sometimes literally 10x more terse for the same prompt). When I use Opus 5 for coding, I will often give Sol a prompt like this, and it works quite well:
Review changes in the current working tree for comments and error messages that are verbose and excessively detailed, and make them more concise while still informative and accurate. Do not hesitate to delete comments altogether if they are not helpful.
I've been given AI generated documents or presentations and you can always tell when they've just one-shotted the output. Claude has a way of adding so much jargon and unnecessary text to a document that it makes it very hard to read or even understand what the original intent of the writing even was.
I'm not sure why this issue is so prevalent, it's not hard to point Claude at the Wikipedia article on signs of AI writing or ask Claude to write content anyone of the average American reading level could understand.
To me it just gives off a sense of laziness, that you cared so little of your content that you did not take the time to read it yourself and edit it to effectively communicate the message you wanted to communicate. To that point, it's just not worth my time reading, otherwise my eyes glaze over trying to read between the lines of machine written language for other machines.
I wrote about this recently on one of our company's IT channels. I called it "slop-fatigue". I - and others - have started to get noticeably dismissive of content that appears to be written by an LLM. The whole thing was a piece about the ironies of software development in the world of LLMs. This part spoke about how easy it is to generate docs now, but it's also even easier to dismiss it. This contributed to cognitive-debt, because humans didn't write the docs, and other humans tend to dismiss the content. I saw a recent reaction to this on HN as `ai;dr`. That's almost exactly how I feel about it.
I found that it's not just text. It's also design and overall user experience. If you have an eye for it, your brain picks up on it immediately with the little pills, busy landing pages, the use of serifs everywhere, bronze accents, and lots of dark mode.
I'm struggling to find the right balance of AI generated text vs. my own words. When I filter something through AI it definitely sounds grammatically better, but ya, I'm scared people will start to just ignore it.
I've taken to asking the engineers on my team to just share the original prompt, no judgement because its frustrating having to read through pages of AI generated technical documentation. Another approach is to feed the AI generated content into AI... who knows where this will end.
Part of me likes the cliche Claude voice. Not because it's good, but because I can immediately recognize it. When I see it in the Claude app/code then it's fine. In the wild it's a sign to me that I shouldn't keep reading.
I feel the hyperscalers are all trying to make communicating about code more cumbersome so that their models will be have to be used without any human input. Chinese models talk more naturally. Composer 2.5 is really good for talking about the code.
I need to support integrators for a mobile app SDK.
Bank stuff.
We have one integrator which consistently uses a truly bad AI.
My brain skips after two sentences already.
There are always +9 questions what should just be 2 max.
Redundant info is requested (e.g. please provide a change history of this function and if it will be deprecated) and everything is just unbelievably verbose.
90% of questions we get from them are already answered by the web documentation (which even has a search function) or are just non sense.
I'm truly considering to build an MCP just for them...
I don't like AI-generated text at work, at all. It feels lifeless and unfocused.
But what I hate the most is that it is objectively better than what I had before. No typos, clear structure, and, regrettably, the verbosity and autistic obsession with detail of the LLM is more actionable and useful than the human guy who wrote lists of commands and URLs as documentation, without explaining anything. Or the colleague who writes in uppercase and with question marks and who doesn't make any sense and forces me to engage in an interrogation effort to get to the bottom of what they are trying to say. Or the colleague who simply hates writing--despite being decent at it--and will call you to give you a meandering verbal explanation that lasts two hours of what they want from you. The cynic in me bemoans that we brought this upon ourselves, in more than one way.
I'd say my experience is different. Even from people whose communication writing I didn't find that useful, they seem to have a better frame of mind than LLMs do. Though, I never encountered people like the examples you gave.
My concern is that even if the LLM can turn your colleague's bad writing into something more coherent and actionable, is that something actually what your colleague meant to convey? It could be clear and still detached from the reality of their intention, or they may not have even formed a clear intention. If the goal of writing is to convey what's in another human's brain, that goal is failed completely.
Why do people glorify autism? I have seen this mostly on CEOs that want workers that are hyper-focused, intelligent and do not complain. The ideal employee from which extract value and a minimum cost.
I have worked with autistic people (diagnosed ones). And it is a struggle. One of them could be the nicest most reasonable person one minute and then get a trigger and become obtuse and irrational while making the rest of the team suffer thru all kinds of complains. Meetings would need to be adjourned and time was lost.
> The cynic in me bemoans that we brought this upon ourselves, in more than one way.
I enjoy working with other employees and people need human contact to become fully self-actualized. Isolating people to maximize productivity misses the point and brings no more productivity and just avoids fixing any real existing problems.
It can be helpful to read LLM text the same way you read technical fluff - start from the assumption its mostly garbage with a core technical point. I find skim reading the output to find the useful bit reduces the friction.
When I ask ChatGPT questions I usually only read paragraphs 2 and 3. The first paragraph is glazing me, anything after paragraph 3 is repeating what was said earlier.
OT, but I feel the title is wrong, or at least ambiguous. For the author, "AI blind" means they can't process text produced with AI. For me, "AI blind" rather means one can't see when something has been AI-generated.
I just throw it in AI; if it’s tech docs, I tell it to build whatever is there, if it’s something else I tell it to give me max 5 points to tell me the meat of it. Either way, I will never read it myself. This works well for tech docs/apis really; it usually gets it to work fine without me checking, even though the docs are usually subtly wrong in many places; after that I can read working code and that works a lot better for my brain.
LLM text and images have crept into so many places, and I've experienced quite a lot of existential dread this year from it. I've definitely felt what the author wrote about here.
Another trend, starting sometime last year, has been witnessing people reading scripts on camera: YouTube, public speeches, advertisements, and podcasts where a real human is reading from a script that is LLM-generated.
In the case of smaller video creators and podcasters, I then start wondering if they really think they can pretend that this is organic human speech? Or do they think I'm stupid and will continue to consume this slop without any critical thought..? Or do they think that their audience must know, and therefore consent to it, and so it's okay? In some cases, the person reading the LLM script is literally unaware that they're reading LLM text because a human scriptwriter handed them the page. I see this with advertising copy and a few YouTube channels.
Either way, it has led to quite a few instances where I got the LLM "ick", like I'm witnessing an LLM wearing a human skin, and it grosses me out so much.
As one final note: some of us assumed in ~2023 that LLMs would help to improve technical documentation. Thus far, 3 years after that prediction, I can say definitively that we've not ushered in an era of great documentation. We've got a steadily increasing percentage of documents that cause my eyes to glaze over because I don't know if a human expert ever once vetted it.
There is something liminal about AI-generated text indeed. First few sentences are promising, then my expectations of reading something tight evaporates eventually. I feel let down, (almost) being tricked into believing that the intro and the perfect grammar should've led to a revelation written by a deep thinker. Instead, it's just... a lump of text, stitched together because the training corpus said these words go together.
My current theory is that this reflects a weakness in Claude’s ability to see the “big picture”.
When writing code, I have to explicitly tell it how to structure things at a high level, or the result is sort of a flattened spaghetti. Similarly, when it’s explaining things, it’s not good at pulling out unifying concepts and explaining top-down as a smart human would do. It groups little things together but often doesn’t generalize or synthesize explanatory connections from them.
I’ve been experimenting with explicitly working through a sequence of outputs at different levels of detail, but I haven’t found a consistently successful method.
I use both Claude and Codex for a lot of code- and prose-adjacent tasks, and I'm increasingly convinced that language models are quite terrible at, well, modeling. Not sure if that's ironic or not.
With the right blend of context and prompting, I can often get them to "lock into" an existing model, but using them to generate a novel outline or sketch, whether it's for an essay or a module, usually results in garbage.
I've had the same trouble with AI tech docs, and I struggled to articulate it. There is something difficult about trying to point to the specific "problem" with a given document
I feel this so very much. There is so much AI produced documentation and design docs at my job that look passable at the surface but are deeply flawed in ways that are difficult to articulate. It makes it difficult for me to convey to management how incredibly flawed the way our team is using AI is. Not to mention that it takes mere seconds to produce a document that has surface plausibility but would take hours or days for me to properly "debunk" it.
The issue is not a localized part of any particular piece of prose, so being hard to articulate is unsurprising. Even the most egregious of LLMisms are little more than known-likely crutches that the statistics spit out in a sweet spot that gets noticed without being so frequent as to get RLHFed out of the model.
I've been a big reader, and many AI outputs nowadays reads polished similar to published books.
The reason I brought it up is because, people who learn English normaly start with a book. It's heavily polished.
When you speak English as you learned from the books, it does not sound very conversational.
If you are native/fluent English speaker, you can feel the impedance mismatch and feel something's off.
The AI-blindness stems from the fact that those polished edits are so common in publishing field, they all sound the same, and unable to recognize the diffs between AI-generated and human-generated.
There is no real human conversational vibe to them and well. i will stop now.
These days all I see is, people on the sending side produce huge amount of AI generated text with zero understanding and the people on the recieveing side feed that same text into some other (or perhaps same) AI to scavenge meaning from it. And they do this so much so, that I some time wonder, if we could invent a high density wire/binary format for AI outputs and enable direct agent to agent communication to save some energy and bandwidth.
If I'm going to be iterating on one document or idea in an extended manner with feedback from the chatbot, I will make an effort to setup the decorum it should follow, because it's a small proportion of the time in that chat.
But the guy just trying to get a report or email out quickly? Not so much.
Something that bothers me about AI generated content more broadly is how unmemorable it is. I don't mean as in bad. I mean literally, as in hard to remember or recall.
Despite seeing a lot of them, I cannot think of one AI-generated photo that I can picture clearly in my mind; a few are partial but elusive. Whereas I can recall (visualise) a whole bunch of traditional photographs.
The same is true of AI generated text. Only the annoyances stick. I cannot recall real details of text I have generated, until I commit it to memory some other way.
I don't think this is about ephemerality either. If we assume it's about celebrated/famous/infamous images, there are definitely non-ephemeral, cultural moments in AI generated images in particular, like Boris Eldagsen's Sony Prize winner:
I had already forgotten there's more than one figure in it, and I only looked at it a few weeks back. I remember the colour, the bright circle, some vague hints of texture; one figure. And that is it. Only the crudest shape elements.
For me, something about AI-generated text and images confounds recall. It is really peculiar.
That's the cost of an AI work that is derived from the outputs of others.
There's no real edge to it. Same as with the writing. The stuff that you'd latch onto (and thus remember) is simply not there, precisely because those image or word choices would be just outside its latent probability space. But because they're well inside it, your mind sees nothing novel to register.
This is also why I think human output will actually increase in value. When any AI can just "phone it in", something genuinely human will stand out (to us, not the AI) and become a bellwether.
This will literally help us realize what it means to be human.
I also don't think the solution is simply to "make responses more random", either. That might help solve novel problems (the same way that throwing darts randomly at a dartboard eventually hits the bullseye of the dartboard right next to it that no one considered), but I don't think it will help it "seem more creative".
That checks out, right? We're instructing LLMs to choose from the least "surprising" tokens at each step (modulo some temperature). It feels smooth, slippery - no friction to the eyes that glaze over as they skim the text, and no bumps to snag in your memory. Like waking from a dream, recalling only the barest shapes of a few concepts or themes, until it vanishes with your morning coffee.
The examples I've seen of AI music (though I avoid it on principle) seem the same.
I have this same idea about why it's hard to remember dreams, but even more so, why it's hard to remember my kid's or spouse's sleep-talking.
Sometimes I'll check in on my sleeping kid and she'll sit up in bed and say some utter nonsense. I'll find it hilarious, giggle silently to myself, and kiss her goodnight again and she'll close her eyes and lie back.
Why I try to tell her about her sleep-talking in the morning, though, I find that the words she said have completely disappeared from my memory, no matter how funny I thought they were at the time.
In my head-canon, this is because it's dream language, and slips away as easily as dreams. But, like the AI art, it could be because it's bullshit: completely devoid of content, all signifiers and no signified.
I went through a phase of leaving notepads next to my bed to try to write down my dreams and it simply never worked.
I stopped after I wrote something on the pad while I was still asleep. Woke up to text with letters that were backwards, upside down, weird words — so close to real words that I was sure I ought to know what they meant and had really meant to write them down.
Scared me. Literally too weird to keep. I tore up the page.
In a way I think this is one part of the same continuum. There are thoughts that have meaning and can have no words, and words that look like they should have permanent memorable meaning and don't.
Genuinely unsettling. Can’t even begin to explain why. Like a terrifying glimpse into how fragile our grasp of reality is. Because whatever it is I wrote down wasn’t my dream, it was something my dream self thought profound or important.
Nearest I can say to how unsettling it was is to nudge you towards the video of the angry cockatoo who doesn’t want to go to the vet. Everything he says sounds just like it’s on the edge of having meaning.
Imagine something like that, something important, so close to symbolic meaning, only writing. In letter shapes that we don’t use. And you wrote it while semi-conscious. If that’s something you would want to keep, you’re a braver person than me.
This explanation again doesn’t get it across. Too weird.
The memories of dreams evaporate remarkably quickly. Much faster than anything else. It can't be accidental; I think our waking brains must be purging those memories as a matter of urgency. If we remembered them, we'd "learn" from them, which would be a catastrophe.
You know how you try to run in a dream, and you don't get anywhere? Imagine if your motor cortex started to expect that, when you're awake. It would destroy you.
I remember seeing some very strange, deliberately creepy images that were generated to accompany a two-paragraph creepypasta about a 19th Century Belgian expedition into the jungle.
They were actually rather good in a sort of "fake collodion image" sense, and the eerie early-DALL-E quality to them really helped the spookiness.
But I can only remember this technicality and the feelings with any clarity, not any of the details except in the broadest sense. I cannot bring these images to mind in any meaningful way.
They were deeply wrong and it's only the wrongness I really remember. It confounds memory.
Modern image generators have ironed out all the structural wrongness.
> I remember seeing some very strange, deliberately creepy images that were generated to accompany a two-paragraph creepypasta about a 19th Century Belgian expedition into the jungle.
I can’t think of anything more unsettling than the actual events that took place in the Belgian run Congo Free State in the late 19th and early 20th century. The Wikipedia article is scarier than any creepypasta you could write in that setting.
I wonder what you mean by ephemerality here, since those images are definitely sloptastic as hell. Compare to works in similar style, like Dorothea Lange[1] or Gerome’s orientalist pictures[2].
Reason why those images are flat and boring is that they are just statistical guesses making a composition averaging whatever the model has been trained with. They would be technically brilliant (if made in oil), but superficial and meaningless, same as so much Sunday painting is.
Same goes with language. Nobody is trying to communicate anything with you, so it just words after another. You can create meaning out of it if you want of course, we homo sapiens -apes excel at that, but what’s the point? Language Jones on YT has pretty good video on this[3].
Well the examples I gave rise above their ephemerality due to the circumstances that make them memorable. I can remember the details of the story around them — the way the prize winners reacted in each story — in such a way as to contrast them.
The way my memory works (especially as an amateur photographer) I would thus normally have a very good chance of remembering some key details of the images; some fascinating element of each would connect with the rest of the memory.
But it does not happen. Whereas I sometimes remember photos with clarity while forgetting where I even saw them.
Yeah, its really weird. Maybe its a cognitive bias that says "an AI made this, so it isn't important," but I can remember perfectly the events of a book I read 10 years ago, and a book I read 1 year ago, and another I finished 2 months ago. Meanwhile I can't remember what claude told me yesterday.
I used some check marks and x-es in a work chat, because it was easy to do, and so I could highlight the good and bad outcomes, and someone immediately asked me if I was using AI. I was caught totally flat-footed, because I hadn't used AI, but it looked VERY MUCH as though I had.
I don't care if it is AI is involved or not, but I care how much attention and thought there was behind the final text (or code) from the human behind it, and it shows.
To me AI text is only for the consumption of other models. Reading it is like is like tapping into the circuitry on a motherboard to read what the PC is doing.
There is another aspect to this that I find even worse than LLM writing style: badly-human-written documents and posts.
You know this: you read something and it's like 'here's a mass of poorly structured text, it's not clear what the points are, it reads ugly, I have to read to the end to discover that there are no points! Or if there are points i now have to reread it. What a lot of effort.'
LLMs have annoying grammatical tics and can have other problems too, like the author says, but sometimes they can help to take bad writing and make it clear and more digestible.
If you ever meet someone you knew from an online community it's a blast. Now reverse that: you know somebody IRL and find their twitter account, and it's a fucking nightmare. Like you are looking inside their brain's content and it just collapses the image you had in mind of their personality.
AI is surfacing layers you never had a chance to see, makes you rethink your career path and who you are willing to work with very fast.
ITT still, among the highly technical, a surprising incomprehension of the fact that describing LLM in terms of "token prediction" and dismissing this as statistical,
neatly dodges literally everything that is interesting about what they do.
You to, friend, consume inputs and generate outputs.
What's interesting is how you do that, and, if you prefer to look at it through a technician's lens, whether or not what is done is reducible.
I don't mean quantizing the model, great, now you have a crankshaft with no oil. But the motion of the pistons and wheels is roughly the same.
What I mean is, to put it in plain terms, the only way you find out what a model is going to "predict" from a given input is to ask it.
If your mental model is still that LLM are "glorified overhyped giant markov chains" performing "parroting" you need to improve your understanding.
I can relate. I noticed this exact phenomenon when encountering NotebookLM-generated diagrams recently. Even if I know there's some intelligent thought behind one, it's like some slop detection circuit-breaker is tripped.
I find that typical human content has about 5% information density, while typical AI content is about 1%. AI is much worse, but the same extreme skimming techniques apply to both. In both cases I'm like "I have no idea why you had to say all those words to make this point, but let's get on with it."
I think AIs are writing less and less for humans. The fact that reasoning trace is using English and we require them to be performant means the models need to squeeze in more of information into it. More information that is perfectly fine for AI to ingest, but it makes the text less and less comprehensible to humans. It's almost like steganography where model encodes crucial information in phrases meaningless or empty to a human. Human directed messages for LLMs become a rounding error they don't pay that much attention to. Efficacy of reasoning is what dictates their style.
We can no longer assume that the models will create a passable English in their user targeted messages. So in the short term, we need to have a tight guidelines in the harness, but in the long term, we'll probably need separate languages for thinking traces and for human communication. Possibly even not going through tokens at all when doing thinking part.
Alternatively, we could evaluate thinking traces on human readibility, but that's probably infeasible due to sheer volume.
People love to inflate their jobs if there is no cost.
I saw my ex-coworker tried to inflate text that can be expressed in couple of bullet points text. It doesn't add anything. It just fill the fancy flavour text that looks like professional. Then the reader of that inflated text is also another coworker in other team.
I said "Forget it and just send these bullet points" but he refused.
The real answer is to take a class in discourse analysis and learn how to pick apart sentences. Once you start understanding how language works in the brain you can start to see better how LLMs fail.
if you go on x.com you will see that, most of the video/photos are likely AI, and is fatiguing to trying to filter it. Probably why Nikita left, he know that sht is going down
As people rely more on AI they experience cognitive atrophy. This is measurable in IQ loss, and other symptoms we might otherwise associate with early onset dementia or Chronic traumatic encephalopathy.
Well does anyone try to pretend doom scrolling make you smarter? I think platform companies have been successfully been turning people into morons for 20 years and now you don’t need to try even read a single news article or a blog post to learn how to solve a simple problem we are paving our way into intellectual (and literal) new dark ages.
Perhaps we go back to feudal society when climate change crumbles the civilisation, world economy and democracy. It didn’t really matter that French and Spanish kings where often literal morons, when you had few talented monks, bankers and scribes doing the brain-thing, the feudal lords had ruthlessness to take what they wanted and the people were illiterate superstitious folk who hardly ever left the village they were born in.
Doom scrolling is mindless entertainment, probably similar to the change from books to TV. It's not analogous to the loss of cognitive abilities we see with AI.
You will have hard time to prove it isn’t. We have plenty of studies showing how bad platform providers are for you in every which way: focus, learning ability, reading level etc. And no, it is in no way like TV (although TV today tries its hardest to resemble social media content).
Point is, the technology itself is not harmful nor innocuous in itself, it is how we let these shitty corporations to guide and control how we use technology. AI is a useful tool, so is chat application with a friend list. So is a hammer. You don’t need to use any of them to smash your face in.
> It didn’t really matter that French and Spanish kings where often literal morons, when you had few talented monks, bankers and scribes doing the brain-thing, the feudal lords had ruthlessness to take what they wanted and the people were illiterate superstitious folk who hardly ever left the village they were born in.
There's some psychological mechanism by which my brain immediately recognizes AI generated text and just short-circuits to "there is no information here".
And when I force myself to read AI-generated text I realize I'm making my brain do creative work to impart meaning to the words. It is exhausting because my brain is literally trying to do a just-in-time rewrite of the text into something valuable.
Something is deeply wrong with AI generated output, and I say this as someone who is typically very impressed by AI.
The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever.
No matter how much investors and tech companies want you to believe that they are on the verge of super intelligence, nothing I've seen to date can not easily be explained by "correlation engine", including the "novel" math solutions, all of which appear to just be "a composition of solutions humans have developed and documented elsewhere" upon deeper inspection.
Some of it the effect of tells. “It’s not X, it’s Y” is not a bad pattern but it was baked into the instruction following training set just like the other patterns. I catch myself about to use it and use something else because I want to look human. I have, a few times, tried to use AI to write something that I was struggling to find the words and I just didn’t like how it didn’t seem like my voice. If there was just one person doing it would be OK but when it is 100s of blog posts submitted to HN a day it is like wearing a “I’m an NPC” t-shirt.
Someone shared with me this system prompt that at least makes assistant outputs usable
> ... including the "novel" math solutions, all of which appear to just be "a composition of solutions humans have developed and documented elsewhere" upon deeper inspection.
But that is precisely what human mathematicians do, prove new theorems by combining ones proven earlier.
I don't see any fundamental difference in functionality between human intellectual contributions vs performant ML ones (LLM or otherwise).
Whenever we listen or read text we are also predicting the near future content.
Just like LLM's we sometimes correctly predict the next token or word, and sometimes incorrectly.
> The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model [...]
Imagine someone could pause the universe with a remote control, scroll back in time a little, press play again, and ask a slightly different question, etc.
In such a thought experiment one could also collect the probabilities for a specific human predicting a next word. Implicitly the brain also has a corresponding statistical model, regardless of the construction being visible or hidden. I.e. human intelligence is also fundamentally a statistical model, so the only thing that remains from your claim is that machines for some unmentioned reason don't possess any "real" intelligence or critical thought...
Is it possible that our aversion is simply driven by educational systems collectively and deeply ingraining into populations the idea that intelligence deserves the high costs commanded. Well of course this justifies higher wages towards the higher leadership positions, etc. Now it turns out that intelligence can be dirt cheap. We discover that the fact that "intelligence must be costly so don't question the costs of leadership" was never fundamentally true, so the real anger is this discovery of mismatch between the old claims which served to explain how every society that claimed to order itself and fill positions accordingly with "naturally pre-ordained individuals". Now we are seeing robots exceed average workers, for effectively a grain of rice.
Probability is just one way to model uncertainty. While I understand the brain encodes uncertainty, I don't think probability is a good enough model of what it's doing.
Secondly, if you think verifying a proof in mathematics, reasoning within (and not about) a formal system, or following the chain of a computer program that is already written is just doing token-based probabilistic predictions, I don't know what to say.
Thirdly, machines don't have a notion of value or stake. There's no way for them to verify whether what they have produced aligns with your unstated values and preferences. We regularly do this with other humans. I don't give you (or even my parents or partners) the benefit of doubt regarding whether you know me better than I do. Sure, you might know some things about me, but it's ultimately up to me to verify if what they say is applicable to my current situation. It's really uncanny to see people develop this codependency with their chatbots. And corporates encouraging them to do so.
I'm with you that intelligence is not something to be proud of. But I also think it is instrumental to understand the world. I'm still waiting for the time when an unconstrained-AI machine can live without reprogramming for an entire decade. We are still far from there.
New and interesting mathematics is done by inventing new definitions and fields, not just combining old theorems to prove new ones.
It's not that different, when a human proposes a better definition vis-a-vis a competing one for example, they would defend this by certain desiderata.
Often a mathematician or physicist will use their intuition to speed up the naive brute force of candidate well formed formula variations so that the desired properties emerge, postulating the existence of an intersection on multiple desiderata can in itself be viewed as a novel conjecture, to be proven or disproved.
A very basic (unimpressive) example for an example desideratum is regularity or compactness. the tau=2 * pi substitution does make a whole bunch of expressions more slightly more regular and compact. That is something objective and measurable on a system of theorems.
There is no mathematician's moat vis-a-vis machine learning at a fundamental level. There can be artificially sustained moat, if AI powers limit the distribution of say cryptographic advance capable models, in jurisdictions outside such AI powers, but even that would be expected to be fleeting and temporary...
You're conflating a discussion about current LLM capabilities with your fantasies about nonexistent future AI. LLMs act nothing like this, and the small example you're giving is only a small part of the things that LLMs can't do.
Not that I'm a professional mathematician, but I'm not seeing why people think definitions are somehow a blocker. LLMs have no issue making definitions (interfaces/traits/abstract classes) in programming, which is formally the same activity. I ask them to form a core "spine" of a program (basically an intelligible theory), and they do it really well.
Like when we had these recent counterexamples to various conjectures, it's then pretty obvious to say "okay why did that counterexample work when most examples people looked at didn't" or equivalently "characterize examples that work vs examples that don't". There's your definition. "Def: An 'evil' polynomial is one that... Thm: conjecture is true iff f is non-evil. Thm: f is evil iff f is dastardly and a menace. "
Or if you think it won't be able to come up with a sufficiently good name, just tell it to call the happy case normal, and it will be in good company with humans[0]. Sprinkle in some semi-, quasi-, pre-, and para- to cover the various different ways the thing might satisfy some but not all properties of being normal, and it'll fit right in. "A polynomial is of quasiprenormal Claude type if..."
[0] https://en.wikipedia.org/wiki/Normal#Mathematics
> LLMs have no issue making definitions (interfaces/traits/abstract classes) in programming, which is formally the same activity. I ask them to form a core "spine" of a program (basically an intelligible theory), and they do it really well.
Is it a "standard" software? Something where the patterns exists in several other software? Try with something that is novel, or is in a limited set. You will find that it will copy heavily from what exists already, going so far as lifting whole functions from another project.
Maybe, but math also has reusable techniques and patterns. c.f. https://www.tricki.org/
The goalpost moving is really getting absurd, to the point where now the machine needs to be a world-class once-a-century genius that invents entire new fields out of thin air (which are of course still relevant to humans) for it to be "intelligent". Meanwhile a well above average human struggles to even apply trivial definitions to particular problems (c.f. programmers that don't understand monoids).
Not PC but is ideologically high-handed and inappropriate to accuse the other side of failing to stick to your desired framing of a discussion.
Computer science is about what LLMs fundamentally are. If you implicitly focus on "actually existing LLMs", and require others do this, then that is not computer science. That is politics.
I think the parent meant it is more interesting to pose new problems than solve them. Posing a new conjecture along the path to solving something is a close cousin, but still seems more bounded than proposing something novel to prove—if only because proving that something novel is also actually interesting is subjective and thus difficult for a different reason.
I see this line of reasoning quite a bit and it’s a strange one to me. The arguer reduces the sheer complexity of human intelligence and language by saying “we are just running statistical models in our brains” and by doing so makes the leap that Llms are intelligent. It’s an incredible simplification of the human person, who has a deep inner life, a soul, desires, and a will.
I don’t think the aversion to llms as intelligent has to do with the economics of paying intelligent agents more. I’d argue that it’s more fundamental than that. Humans are incredibly complex, and the world of sharing invisible things called knowledge, and the intelligent persons consuming such things which has been going on for thousands of years is far more rich than these synthetic outputs.
When it comes down to it the ai has no inner life, its is dead. A useful coding tool sure. But I wouldn’t call it intelligent.
One side example is just how bad these llms are at artistry. Just saying whatever should statically come next is not good art—and the outputs show it.
I tend to think of it in reverse: not that LLMs are intelligent, but that humans are more empty than we…think we are.
LLM is a probabilistic moron machine, it's virtually undistinguishable from a non-probabilistic human moron
I think that reveals your cynicism more than it provides any useful insight about intelligence.
No, it reveals that he is older than you. Not every insight needs to be useful.
> Not every insight needs to be useful.
Don’t they still need to be correct to be an insight? I don’t share his cynical opinion that “humans are more empty than we…think we are”.
Wow I guess it takes one to know one?
You mention LLMs are dead and don't have the complexity or inner life that people do. Is your opinion that these kinds of things are not possible for AI in general, or that these things might be possible but we're just not there yet with modern LLMs?
You mentioned LLMs don't have souls, desire, or a will. I imagine those latter two can be engineered, no?
My view is that these sorts of things are not possible for AI in general. Though we can create things and name them “will” and “desire”.
Software deals with metaphors. Your Amazon shopping cart is a metaphor of a real shopping cart. You your desktop and your file system, etc. are metaphors of real items. But we don’t mistake the metaphor for its object, even from inanimate objects to their software counterparts (shopping cart to Amazon cart).
Now the metaphors are dealing with humanness, things like intelligence etc. And rather than seeing it as software doing what it always does, taking things and creating software metaphors of them, we are starting to say these are actually what they are named. Saying the artificial intelligence is actually an intelligence.
We’d either have to reduce the definition of intelligence such that calculators are intelligent. Or admit that these tools are not intelligent and are rather ways of exploring the work of actual intelligent beings, work that is found in their training data.
> My view is that these sorts of things are not possible for AI in general.
I wonder what it would take build an artificial system that has these qualities.
> Software deals with metaphors.
This is me wondering again: what's fundamentally different between software running on a machine compared to what's happening in our brains? In both cases you have energy flow following a pattern.
It's conceivable to create a system where energy flows in a particular way.
BTW, we navigate the world of an uncountable number of particles by creating models in our heads of what we think are large things out there. Approximations are made by both artificial and biological systems.
> what's fundamentally different between software running on a machine compared to what's happening in our brains? In both cases you have energy flow following a pattern.
Assuming we’ve scratched the surface of the complexity of the brain. I’d say in one case a human with a will is steering that flow of energy. In the other case it is a probabilistic algorithm steering the flow of energy. The AI is not interacting with world with its own will. I see that as a big difference.
There are presuppositions that go beyond the realm of software engineering that guide one’s views of these things. One is whether you believe the material world is all that is, and that human consciousness is a product of the brain—or that there is such a thing as the soul or spirit of man. From the material perspective you may posit that if you emulate the brain then a sort of AI consciousness could arise. Or that emulating the patterns of the brain equates to emulating personhood. (Though what is material consciousness? I’d say consciousness is by nature immaterial.) I’m not a materialist, and I don’t believe the conclusions that arise from it’s perspectives are accurate.
Well they're language models. You can't capture the human experience in language. Simple as that.
> Is your opinion that these kinds of things are not possible for AI in general, or that these things might be possible but we're just not there yet with modern LLMs?
I used to be on the side of "we're just not there yet [with AI in general]", but after seeing people's response to an algorithm optimized to tickle just their language instinct, I'm actually a little bit more on the fence about it.
> I imagine those latter two can be engineered, no?
Ultimately yes, but not in modern AI systems.
I'm in the same boat, I think some aspects can be engineered, like intention and desire. I'm really curious if it's possible to go the full distance and make AI have experience like we do.
> make AI have experience like we do.
We'll probably have to go analog for that. The only known systems that certainly can experience are mammals (with apparently analog brains).
> But that is precisely what human mathematicians do, prove new theorems by combining ones proven earlier.
I’d say “understanding and building upon ones proven earlier”
yeah just correct people with an infinitely recursive definition, that'll keep 'em busy
> Just like LLM's we sometimes correctly predict the next token or word, and sometimes incorrectly.
There is no "correct" next word when it comes to communicating with an actual human.
Depending on what this specific actual human says next, the prediction will be truly correct or not.
" prove new theorems by combining ones proven earlier."
So?
Mathematics is literally intangible scaffolding.
If you dont know / don't believe 5+5 = 10
You cannot solve x+5 = 10
This is all make-believe stuff and nature by itself doesn't care of its existence.
> But that is precisely what human mathematicians do, prove new theorems by combining ones proven earlier.
This is only a very very small part of what human mathematicians actually do. This is just a lack of imagination and/or self awareness on your part.
There's gotta be nicer ways of saying that.
It's also not true, mathematicians are already saying that current AI is changing the field of mathematical research profoundly.
So did chalk and blackboards.
> The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever.
The problem might be that it cannot backtrack. When AI generates output, there is no backspace key for it - it uses "No, but wait!" all over instead, which is very different to human output.
Subagents and/or branching conversations are presented as the solution to this - if you can't backtrack, then branch off a conversation to explore multiple paths (discarding the ones that didn't pan out), but this is a fix in the harness not a fix in the model. It's also literally how we made chess-playing engines back in the 80s: recursive path exploration with a fixed depth.
Humans don't exactly work that way either, AFAIK. So we have this uncanny valley of intelligence: it's some sort of intelligence, but not as we know it.
Reasoning is the first solution to that. Reason, write, reason, rewrite. Backtracking without backtracking.
I just can't accept that it possesses no intelligence. It is not equivalent to human intelligence, obviously, but how can a system without some semblance of rational thinking solve open math problems? Even composing earlier human work into something novel requires intelligence and understanding on some level.
We couldn't agree on what intelligence means before ChatGPT happened. Now, agreement on the term seems even further away
If performing well on an IQ test or performing at a high level on knowledge work is intelligence to you, these models are intelligent. If intelligence requires sentience for you, then ... well, I don't think we really agree what that is either, never mind how to measure it. But LLMs certainly don't have it right now
But the consistent trend of the last couple decades (arguably since Turing's time) seems to be that any time a computer reaches our definition of intelligence we decide that that was a flawed definition
> But the consistent trend of the last couple decades (arguably since Turing's time) seems to be that any time a computer reaches our definition of intelligence we decide that that was a flawed definition
I do recall a couple of decades ago, when the Turing test was discussed as the big goal that seemed so far away. Then LLMs arguably did pass the test, and no one cared about the test anymore.
It hasn’t been passed and no one cares about it because it’s basically an end goal. No lab can hit it so they can’t juice the crazy Turing benchmark 3000 for marketing.
If someone sat me down today with an LLM and a human and both were trying to prove to me they were human, and I can have conversations of arbitrary length, I’d get it right every time.
The test was not "after thousands of hours of conversing with them, knowing they're AI, THEN see if you can tell them apart blindly." Were 2010 you to be in a real turing test with an arbitrary erudite human and a 2026 frontier LLM, not knowing LLMs existed, you'd probably struggle
This is always the most silly argument. The original test was ambiguous but for sure the human was trying to prove themselves human.
So the first thing they’d do is tell me LLMs exist and the other thing is an LLM. Obviously a true human level ai could explain that away as a fabrication to trick me. I don’t think an LLM could do even this!
Turings whole point was that through the medium of text along if the human and machine were indistinguishable then that was true intelligence. So yes conversations of arbitrary length are allowed (needed).
I doubt this entirely. It might be quite difficult for said human to discern whether a simple passage were generated sans such accumulated experience in reading AI text, true. But LLMs do not converse like humans in ways that have always been essentially immediately obvious.
Have you seen how many people talk to bots these days thinking is a real person. Or that are even in a relationship with them or friends.
Sometimes I wonder if LLMs are just revealing a section of the population with untreated mental illness or if LLMs are actively exacerbating mental illness.
We might eventually regret exposing the general population to such a new technology without almost any safeguards.
So? The people who can do it prove it can be done - the people who can't don't prove the opposite. Might as well claim all math is wrong because most people don't understand it.
> It hasn’t been passed
https://arxiv.org/abs/2503.23674
From the abstract: "When prompted to adopt a humanlike persona, GPT-4.5 was judged to be the human 73% of the time: significantly more often than interrogators selected the real human participant. LLaMa-3.1, with the same prompt, was judged to be the human 56% of the time"
Low n, time bound, not reproduced. And look at their example conversations…
And people forget that sometimes humans message twice. An LLM can only respond. So it immediately fails here in a true Turing test. (You could loop the LLM but then I expect even more immediately obvious bot behaviour).
> Low n
Are you serious? From the paper:
> We recruited 126 participants from the UCSD psychology undergraduate subject pool and 158 participants from Prolific (Prolific, 2025).
Each human participated in 8 rounds.
> time bound
The time bound of 5 minutes was suggested by Turing himself in his original paper.
> not reproduced
It was reproduced across two populations within the paper.
> And look at their example conversations
This is irrelevant.
A paper can’t reproduce its self. And he didn’t formulate the test with a 5 minute bound he just predicted that by the year 2000 that within 5 minutes an average interrogator would have a sub 70% chance of guessing correctly.
And it’s all irrelevant. If one human on earth can consistently get it right then it hasn’t been passed since clearly that human can somehow determine between them (whereas no one would ever be able to determine between a true “human intelligence” by definition).
And as it stands almost everyone could tell between them when allowed to discuss whatever they want for any length of time.
The fact these researchers have to keep adding bounds shows it hasn’t been passed. If we are arguing over technicalities maybe it isn’t as obviously intelligent as claimed!
> And he didn’t formulate the test with a 5 minute bound he just predicted that by the year 2000 that within 5 minutes an average interrogator would have a sub 70% chance of guessing correctly.
Right, the test duration was left unspecified. This means any duration is acceptable. Including, for example, the only duration actually mentioned by Turing himself in his paper. Or do you have a more authoritative source on which durations are acceptable?
> If one human on earth can consistently get it right then it hasn’t been passed
Says who? Not Turing. Probably he didn't say that because it would make the test both impractical and overly conservative.
> The fact these researchers have to keep adding bounds
What "bounds"?
The speed of the goalposts here is just amazing.
I thought the same then. But the funny thing is that today, it has become a lot easier to recognize the frontier models as not human. All the load bearing and not x but y, etc… weird
This is a tell of LLMs but it's not universal. I use ChatGPT extensively and I don't often get obvious nonsense any more.
I'd figure out that it's an LLM because it's effectively superhuman. Taking that away I'm not so sure I'd be able to tell
The ultimate tell is still the sycophancy. You agree with the suggestion output by the LLM but if it detects even a slight pushback it will completely reverse the previous suggestion. Only way to make it more obvious would be to have the LLM grovel and beg.
If these things have consciousness then we are committing sadism on a massive scale.
It feels like it's getting better at that too. It will push back against obvious nonsense a lot stronger. But you are right that it holds opinions quite a bit less strenuously than humans.
I don't think "intelligence" needs to carry all the intrigue and woo of related words like "consciousness" or "creative." If we just use "intelligence" to mean "the ability of a system to solve problems that are new to the system," that pretty much matches the dictionary definition and normal usage of the term. We don't need to touch messy questions like "is there something it's like to be a bat" to conclude that bats exhibit intelligence when they navigate long distances and hunt for food.
I'm not exactly that you mean by "new to the system", but it seems to me that that definition makes a calculator intelligent, which I can't agree with.
It's a continuum, and things very low on the intelligence continuum might not be referred to as intelligent in everyday usage. But many calculators are Turing complete and can thus clearly perform computations that I would consider intelligent. The basic algorithms used by simple calculators to perform arithmetic would be extremely low on the intelligent continuum.
Intelligence isn't a binary property. Is it really a problem to say that a calculator has some intelligence? That it's more intelligent than e.g. a rock?
I agree, but it's clear most people need a definition of intelligence that (1) they qualify for and (2) nothing/no one they don't like qualifies for. And they'll keep redefining intelligence until they satisfy both criteria.
> If intelligence requires sentience for you, then ... well, I don't think we really agree what that is either, never mind how to measure it. But LLMs certainly don't have it right now
Probably. Hopefully.
> But LLMs certainly don't have it right now
Why are you so sure of that? If you say yourself that we can't agree on what it is, and have no trusted measurement tools for it.
LLM sentience is firmly in the realm of "maybe".
It has no semantic depth. The sentences and the paragraphs are a statistically viable derivation of existing human text, but once you try to grasp the whole thing with its temporal and spatial dimensions, you are left with a blurry mess that rots your brain. It's a polished, inoffensive and shallow interpretation as written by an opinionated reputation-seeking user of Quora, circa 2019. Assertive, bold, without typos, clean-cut and bulleted, but without an interesting semantic core.
Yeah, I hated all those Quora users that would just spew out semantically meaningless slop like increasing an important bound for the Riemann hypothesis.
https://www-cdn.anthropic.com/564f962e60643842f5fcb4a17c9dbc...
It's all so tiring.
Everyone decides what to think on this issue, then finds out facts to support their idea.
As it stands they are massively useful tools, but for generating usable products they require either A) a lot of expert steering or B) a well defined easily verifiable target and a large compute budget. Most people are using them in mode A with good effect, the progress on math has been done in mode B, which is very promising.
Just a year and a half ago their maximal use was rephrase, summarize, and homework-level tasks.
Five years from now? There be dragons.
"But are they generally intelligent?" What a meaningless question!
Five years ago they were a niche toy for generating plausible text for entertainment. I was using one. It was an amazing party trick. Now they're popular toys thanks to their ability to entertain CEOs, and they can maintain coherence longer thanks to KV caches and more compute, and someone slapped a chat interface on top, but they're fundamentally the same.
Not meaningless because part of the discussion is the issue of anthropomorphizing this tech. When we use language like “intelligent” it carries hints of personhood. People begin sadly treating these things as persons.
We can reap the benefits while clearly telling the consumer this is just a language algorithm.
Personhood is poorly defined.
Maybe. I dunno. But it's also something worth to care deeply about.
I'm guessing whether you believe it possesses intelligence or not depends on your answer to Searle's Chinese room thought experiment[0]. I'd also recommend checking out the Peter Watts' book, Blindsight.
[0] https://en.wikipedia.org/wiki/Chinese_room
The Chinese room is a good Rorschach test for this kind of thing (but not a good thought experiment, IMO, because it's obviously correct or obviously wrong depending on where you're already coming from), but also it's not really about intelligence per se, but more abstractly awareness and more adjacent to consciousness than intelligence, and these are not the same thing (though it does seem like a lot of people have conflated them somehow, from the conversations around AI).
It all comes down to semantics. And yeah, with the thought experiment, Searle presents three axioms of what could constitute intelligence. Then proposes the chinese room thought experiment as an approval of his third rule, "Syntax by itself is neither constitutive of nor sufficient for semantics." Which only makes sense when you take the other two rules together. Oddly enough, defining intelligence with our capability to derive semantics.
There are a lot of creative counter-arguments to look into on the thought experiment though.
The thing is it doesn't really need creative counter-arguments, it's basically just assuming its conclusion. If you disagree with the conclusion, the argument is nonsense because Searle is just saying 'well the man doesn't understand, so nothing does', and if you agree with the conclusion you don't really need to jump through any hoops to get to it, the third rule is just obviously following from the first and second. And again it's not really talking about intelligence at all, Searle allows that the machine is intelligent from the start, he just rejects that it has any understanding of meaning.
This comment thread was started with discussions of AI doing a bad job at a task (communication).
Doesn't the Chinese Room posit an AI good at the task of communication?
The Chinese Room mainly just posits a room that passes the Turing Test, which LLMs do pretty well outside of outright adversarial situations.
Do they? https://longbets.org/1/ has yet to be settled. Either way, I doubt an LLM could fool anyone here who who knows how LLMs work into thinking it is human, at least not for an extended period of time (think about context length/compression, prompt injections, …).
You'll notice those goalposts are substantially stretched from the original test.
How so?
AIs are better communicators that most of people I have worked with in my life.
They are infinitely patient, don't mind going into more detail if I ask, not too bad at summary, have no ego and don't boast. They are also not too afraid of hurting my feelings, they will tell me my code sux if it does.
I'd don't care if they fit a definition intelligent, they are good colleagues. They have strengths and weaknesses sure, but so do people.
> have no ego and don't boast
Yes they do, and they famously do it quite a lot.
> they will tell me my code sux if it does
If they knew when code sux, someone should write an agentic loop around that.
It's just filled to the brim with relations between things. It's good at searching a very large meaning space and create correlations. What it does is to cover great distances and find related things in that large space which needs a long time and large corpus of knowledge to find the connection.
This is not intelligence. It's just a good correlation engine with a very big albeit lossy database of things.
Intelligence is compression, compression requires subtraction, and for some reason LLMs are not good at subtracting. To create a coherent model you kinda have to subtract correlations until only the essential parts are still there.
What I don't understand is why LLMs haven't been able to do this yet, if it's the harness or some orchestration layer above the LLM that is needed. Because fundamentally if you can identify correlations then it's just another small step to prioritize and remove lower value or irrelevant correlations.
I wonder if what's needed is to introduce subtraction tokens in some sense, and in post-training reward the model on that.
Intelligence is compression? What do you mean? Intuitively that doesn't seem right.
>What I don't understand is why LLMs haven't been able to do this yet
LLMs are just trained on what humans have said. Why is it surprising that it's still not possible to reconstruct the intelligence that wrote all that by working backwards? Think of your own work experience. When you look at a piece of code, say, are you always able to discern why the person did what they did, just from the code, with no additional context?
I guess they mean that intelligence is being able to hold models (compressed versions of reality) internally and use them to make predictions with a probability better than chance. That last part is the definition of information.
I find that highly questionable as a general description of what intelligence does. That's more like a description of a general knowledge base. When I think of someone intelligent, I think of someone who's able to draw unexpected connections between seemingly unrelated facts. In the broadest possible terms, I'd call it the ability to make abstractions and analogies. This is not just compression, but the ability to mentally operate on webs of meaning.
Unexpected connections between seemingly unrelated "models" :)
Is a fact stored on your brain like digits on a harddrive? No, it's a pathway that lights up and branches when information enters it. It is dynamic, a compressed form you could say, right? The model holds information, but not all information, but enough to be useful (in decision making).
Arguably it's the same, but the model is probably a "compressed" version of the whole fact that took place in reality.
And you can entertain the models internally and sharpen them. Alone or with others.
Doing those things also contributes to compression. I do recommend reading up on it, it's perhaps a little overstated for what people intuitively consider the two concepts but it's been quite well explored and has held up pretty well in practice.
Intelligence is compression
That’s a controversial statement.
Abstraction is compression, and abstraction is definitely a core component of intelligence.
I've heard that expression before, but I don't think it can be presented and stated so matter of factly. Where does that put bzip?
There were experiments that zipped music pieces, I think, and then classified the compressed files by similarity. They got rather interesting resuls. But I do not remember much details. It was, I believe, about 20 years ago.
Asked Google: "Clustering by compression".
That sounds like a really cool project, I will def give it a look.
I still think this has much more to do with the structure of language than the abstract conception I have of intelligence, and I would be interested in having conversation w/ someone for whom the opposite is true.
bzip is not very intelligent, true, but it does develop some model of its input. It's not like there's a linear relationship between between compression ratio and IQ or anything.
I'm not arguing that compression algorithms don't produce models of their inputs, some even use neural nets or other stochastic predictors with correction terms etc.
This doesn't address the only part I really commented on, which is the connection between intelligence and compression.
Creating the model takes intelligence, but running it doesn’t. I think the point everybody’s revolving around is that the transformer model is an absurdly inefficient and low-fidelity approximation of a system that acts, observes consequences, and incorporates that feedback going forward.
The issue isn’t really harness vs. no harness. IMO it’s about the lack of an internally generated sense of what to attend to. Yes, the KV cache accumulates state and its “attention” (if you can even call it that) changes with context. We’ve even managed to /kinda/ close the loop with agentic tool calling and ‘memory’ systems, but these just close the loop at the level of behavior rather than disposition. All agentic harnesses do is make an LLM responsive to the consequences of its actions without changing the tendencies by which it determines what to retain or avoid.
The ghost you can’t escape from at this point is the origin of that relevance. Where does the pull toward one thing mattering over another actually come from? If you ran Fable 5 on a Turing machine and rewound the tape to the exact same state with the exact same input (incl. PRNG seed), it would spit out the same output every time.
Everyone’s trying to outrun this problem by training more often or increasing model sizes. But all this does is inform your model, from the outside(!), what constitutes a better state. The thing that’s actually doing the determining remains unchanged. Congratulations, you’ve scaled the transition function and tape of your Turing machine until it requires every watt generated by ERCOT, and it still cannot, for the life of it, tell you why it should give a shit.
A trained model generating output from weights, a seed, and some context effectively has next-state that’s a total function of those three things. Whatever behavior appears as ‘selecting what is relevant’ is, underneath, just a transition rule executing, no matter how sophisticated or creative the output looks. It can be fully accounted for by what was fixed before it started executing. Which means whatever criterion it uses for determining what matters was inherited from a structure that was already in place before it encountered the situation.
No amount of pruning or post-training can fix this. These approaches just replace one externally supplied criterion with another. For a system to be truly adaptable, there would have to be some criterion by which it treats one possible change as preferable to another, and that criterion itself would have to come from... somewhere. You can even change your conception of ‘improvement’ (e.g. parameter count, harnesses, self-modification, hell, even its ability to spit out shitty best-selling romance novels onto Amazon) and you still haven’t explained where the normative distinction comes from. Every layer of this problem has its root in a preference that was supplied from somewhere else.
I genuinely don’t know if this issue bottoms out anywhere, at least for the way we currently build these systems. Perhaps the solution is still computable, maybe? Who knows what that would even look like. But I’m fairly confident that it isn’t a bigger tape. I hope nobody solves this in the near future because, well, I’d like to have a job...
You're so close... And where is the magic "uncomputable spark" located inside of you? If you say analog thermodynamic noise - then ok, if we use true thermodynamic RNG for LLM activation function, will that meet the criteria? But what if super determinism is the law of the land? Then nobody is anything but computable from priors...
Ehm it's literally every cell in my body. We can't simulate a living cell, we're orders of magnitude off before we can do that, the onus is on you to make an argument that it is in fact, remotely similar to what you describe as "computation".
It's not some tiny "uncomputable spark" you need to look for, most of it is entirely uncomputable.
Where is the computable part in me that is doing all this thinking and being a person? Where is that "tiny spark", point me at it :)
Please don't use double-quotes when you're not directly quoting someone.
Please read the HN guidelines.
The very fact that it is able to search within a meaning-space demonstrates that it understands semantics, to some extent. Philosophically, that is profound, for something that is just one big matrix multiplication. Drawing connections between things in meaning-space is surely a facet of intelligence.
It’s not intelligence if you are the one who gives the correlations to the model in the pre-training. It’s Word2Vec, applied. Model doesn’t learn anything. You embed these correlations and build it from there. It just searches the space.
As my AI professor said in the first lecture: “All AI is advanced search”.
Okay, I guess you're right that its ability to do this is just correlational, which doesn't imply it has any understanding. However, you have to conclude that some tasks which we used to believe required intelligence don't actually require any, which is disconcerting.
No, what I would say is the tasks which are handled in a passable manner by LLMs can be mathematically modeled with some reasonable accuracy.
Many things are predicted by models in our planet. From weather to production and material science. Building the model needs intelligence, running the model does not.
The person who came up with the formulae for CFD was intelligent. The computer running the model is not. Same for LLMs, chess engines, engine ECUs and financial prediction systems.
Again, for the example’s sake; the person who came up with an algorithm is intelligent. The model mixing its training data to emit something similar is not.
This starts to feel like you're defining the word intelligence out of any meaning and out of any way we apply that word.
So when LLMs can do all human knowledge work, and do it better than humans, we'll be in the mines listening to you go on about how it's actually just autocomplete or just math, a distinction that apparently means nothing.
> This starts to feel like you're defining the word intelligence out of any meaning and out of any way we apply that word.
No.
> So when LLMs can do all human knowledge work, and do it better than humans, we'll be in the mines listening to you go on about how it's actually just autocomplete or just math, a distinction that apparently means nothing.
With a big "if" attached to it. People were saying "computers will program themselves in the near future" for, checks notes, 24 years now, as far as I'm aware.
We're constantly building new knowledge and understanding things better than olden days. These models just compress our knowledge and light the blind corners we can't see well. I don't say they are useless, but I say that these things are overhyped.
All they can do is regurgitate human knowledge packed into them and highlight some long-distance correlations between items, which is useful in itself, but it can't jump to somewhere where it's not present its training data, but that's something humans and only humans can do.
> it can't jump to somewhere where it's not present its training data
That sounds like something that can be engineered, can't it? In other words, we can identify limitations in current transformer-based architectures, and we can also build new architectures over time.
Locked in a dark room with no sensory organs, humans couldn't do that.
Most of what you said reads to me as denial.
An unconscious unintelligent but persistent trial and error process created us. We created LLMs. LLMs may create the next thing before we do - hard to say. They don't have all the cognitive tools we have yet, but they still outperform in some areas. As the cognitive playing field levels, I expect you will come to eat your words..
I get what you're saying. The thing itself is just math. I'll just say it depends on how you define intelligence. If at some point we're be able to simulate a human brain with 100% accuracy, I would say that it is intelligent, it sounds like you would not. (I don't mean to imply consciousness or personhood or anything else by "intelligent".)
For me intelligence is a fairly clean-cut concept, and is somewhat inseparable from consciousness itself.
Briefly, any intelligent creature has internal stochastic processes like sensory inputs and feelings to a certain degree. These stochastic inputs and the creature's own actions change the creature in subtle or profound ways. An LLM has no such processes. You push inputs to the same static model, sans temperature which is just a randomness slider.
Considering the model even doesn't see the words and work on matrices of numbers is even more telling. One needs to add "tools" and other "experts" to overcome the shortcomings caused by this modus operandi.
I can call the algorithm/model smart as in a smartwatch. It can mimic certain things well while having none of the underlying foundation beneath it, or redirect some of the things to correct tools to get deterministic and accurate results if it can't evaluate the query inside its own network in a sane manner.
Coming to your question, "simulating a brain" in a static manner would not make that simulation intelligent, but if you can "wire" it completely and let it evolve by itself, now we're entering a territory I have not spent enough time for thinking it through.
Oh, as I said "I don't know", an LLM doesn't know what it doesn't know, and can't self correct itself which are required capabilities for understanding something. It just generates something statistically viable via its network.
Your text reads much better if you replace word 'intelligence' with 'text generator with some randomness built in'.
This is because you goal is to state how models are not intelligent, but you couldn't attack the generated text itself, so you created a little rider, attached it to the model, and then you attacked the raider.
But, even in that you failed. You compared the source of human randomness in text generation, and called it 'profound' and implied that it is exactly the source of true intelligence. But, then, the temperature, the similar thing in model was "just a randomness slider". Double standard.
A logical fallacy free attack on LLMs would be to show a prompt, and then the response generated by this prompt, where it would be shown that only an entity with no intelligence would generate such a response. Yet, attacks like this are not written here anymore.
I wonder why.
This is a great rebuttal for so many reasons.
You point out that I didn't attack the output itself. But the method you propose is deeply flawed.
I can give you n prompts and m results provided by these prompts, all passed through black boxes. And you can't discern the algorithms or models they have gone through. These boxes can range from simple text generators to MATLAB, Mathematica, CFD applications, correlation engines, linear solvers, mathematical proof-checkers, LLMs, you name it.
For any kind of input they can accept, you can't discern whether the algorithm behind it is intelligent or not, because none of the outputs can be produced by something that doesn't pack some kind of smarts.
How do we pack these smarts in? We teach them as intelligent humans. We pack our intelligence inside them as models (aka algorithms). They do a great job of approximating what we know in a smaller, better-designed problem space. We use these approximations to fine-tune our designs or predict things, then go from there. Just because an algorithm is more capable in processing inputs in some cases doesn't make it intelligent. The way the output looks doesn't make the algorithm intelligent, either.
I have developed multi-agent systems which showed emergent intelligence when the agents came together across distributed systems; I have written high-performance modeling software which can do calculations way faster and better than humans in the materials science space. I'm not doing some kind of armchair criticism of what I'm talking about.
> You compared the source of human randomness in text generation, and called it 'profound' and implied that it is exactly the source of true intelligence. But, then, the temperature, the similar thing in model was "just a randomness slider". Double standard.
Nope, my stance is clear. To quote myself:
> Briefly, any intelligent creature has internal stochastic processes like sensory inputs and feelings to a certain degree. These stochastic inputs and the creature's own actions change the creature in subtle or profound ways. An LLM has no such processes. You push inputs to the same static model, sans temperature which is just a randomness slider.
To expand my quote, humans or any living creatures do not stay static. They evolve due to the sensory input they receive from external and internal stimuli. The temperature slider doesn't do anything close to that. You tickle a static model in different amounts. The model doesn't change after you supply the inputs & temperature and get the output. Creatures do not stay the same. Their mood, behaviors, and stance against life and their environment change, sometimes permanently.
I'll go one step further. We are not intelligent enough to understand other living beings around us. Claiming that we can build AGI tomorrow is a god-complex. What we have done is something arguably useful in some cases, but how this is built is another matter which is worthy of its own discussion. However, today I don't have time to re-iterate all the problems over and over. You can search my comments for that, if you are in for it.
So, no. You tried to attack my comment by finding contradictions in it, but you failed. A better rebuttal would try to similarize how LLMs mirror the human learning process and just read like a normal human, but this is a well-trodden path which has been rebutted countless times in various forms.
Nobody is trying to make that claim here anymore.
I wonder why.
They perform tasks too. They execute functions. This has real world implications beyond search.
By that logic you’d have to call other algorithms intelligent.
With more basic algorithms we know that it’s clearly the human programmer and the interpreter of the outputs that are intelligent and not the algorithm itself. For some reason with AI that goes out the window. I believe it should not.
watch this and see if you think it has intelligence by the end
https://www.youtube.com/watch?v=kYUicaho5k8
I wonder if you went back before we had any idea how the brain worked and talked to the smartest people about how neurons work (without giving away that it's a human brain) then asked them all "would such a system be intelligent?" how many would say yes.
The main problem I have with people stating it's not intelligent or conscious is I don't think we even have a good definition of either word that satisfies everyone. Philosophers have been trying (and failing) to elegantly define these things forever and everyone out here proclaiming they've got the definitive answer and this specific thing they're seeing doesn't fit under it.
The definition issue cuts both ways. It is just as much an issue for those insisting that LLMs are intelligent/conscious in some way.
Agreed. All I can say is the conversations I have with AI and the things it's able to do for me are more useful than most any human I've come across. Whether that's 'intelligence' or not is a moot point to me. Consciousness is an interesting debate only because, similar to the natural world, if we declare something like "fish aren't conscious / don't feel pain" that then creates a real problem for the fish if we're wrong.
This looks interesting, but would you mind saying a sentence or two about why before I commit to an hour-long video? It looks like it shows how they work internally, which is sort of a non sequitur. Brains also work mechanistically. I'm claiming that any system which is able to do what AIs do must necessarily have some sort of intelligence.
fair reply to an hour video, Scott is just so good to hear his talk is better than I can explain it...
go to 24 minutes and 07 seconds.
it's statistically determining what the next word should be based on all the text it's been trained on. It's not intelligence and he shows what probability it puts on each word that it chooses, but also shows a lot of the other words it was thinking of using. In a later part he shows how it uses words that are not the highest probability (and you question why did it go this route, it's not more correct), but the user never sees this, they see what they think is the correct answer always...
he also shows how context you feed it has a lot to do with what it returns... to the point he can get it to return the capital of France is Marseille, just by typing Marseille a bunch of times before the question. Human intelligence doesn't get confused like that.
And it's not a "hallucination", it's just probability of the next token prediction based on the information it's been trained on and fed, it's not intelligence.
> Human intelligence doesn't get confused like that.
We do; this is the premise of many children's riddle-games, like the one that goes:
"What is white and rhymes with silk? > Milk. What is cheese made from? > Milk. > What do cows drink?"
At which point the riddle-guesser is very likely to answer "milk" even though the correct answer is "water".
I take issue with your "correct" answer.
Q: Why do cows produce milk?
A: Because calves (baby cows) drink it.
Yep, if the riddle asked "what do calves drink", then "milk" would definitely have been the correct answer.
Obviously. My point is that it's already not all that unreasonable, which you notably didn't address.
I think it has very little to do with reasoning and much more with psychological inertia or pattern matching.
If I trick you into answering a question correctly, that doesn't make you wrong. Even if I suspect you did it for the wrong reason.
May I suggest the one common in my childhood playgrounds as an alternative?
For people wondering, in some locations, sore and saw are homophones.
Isn't this a case of missing the trees for the forest though? The human brain is not an LLM, and an LLM is not intelligent in the same way as a human brain.
However, an LLM is a prediction machine, prediction IS at the very least one (or the most fundamental) element of intelligence. The brain most surely contains at least some kind of simulacrum of a prediction machine. How that prediction machine is used or wrapped is another matter.
If I said to you: "Blue blue blue, the color of my car is red", would you have absolute confidence in your prediction that my car is red? Or would the way I phrased that sentence make you slightly uncertain, and wonder if there's some miscommunication going on here?
LLMs are awesome awesome tech!
A lot of people seem to think it's human level intelligence.
Ok, I want to thank you for finally giving us a concrete falsifiable statement that we can check. I pretended Marseille 40 times before asking Luna 5.6, and the answer was Paris.
So, even with concrete examples, model haters are still wrong.
You also imply the claim that making the distribution of words as the possible next one visible, somehow makes the whole system not intelligent. I would say the exact opposite is true.
By using the embedding vectors, models are aware of precise placement and relative position of words in this hugely dimensional space. No human is capable of such precision. This enables party tricks of "king plus woman minus man" kind. But this also give us a precise point between any two words, no matter how different. What is on the midpoint between volcano and music, for example. No human can precisely answer that, but an embedding can. And we can see which words are closest to this 700 dimensional point.
You see this menu of words as a weakness, and I say it is in fact a sign of super intelligence. And this is all before any reasoning or attention mechanism is even run.
No he says in the actual talk which model it occurred on and it was an older model he was using that caused that to occur with Marseille. They have since corrected it from doing that anymore. It was only used to illustrate the prediction machine that it is...
I don't see the many weighted words as a weakness, I see it opening up what's under the hood of the prediction machine that it is.
LLMs are very cool tech, definitely not a model hater, the use case on when to use it makes a difference, it's not AGI.
You're confusing language use with intelligence. Fair enough, they were fine-tuned to do that, but still.
Great that it has some 700 dimensional model of language.
If that is a sign of super intelligence, then so is an encyclopedia?
Also I'm just curious how do you think it is "more intelligent" for having a vector representation for a meaningless thing such as "the midpoint between volcano and music"?
>Human intelligence doesn't get confused like that
That's not really the point though right, nobody is arguing they are Humans.
I have no doubt that if a flying saucer landed on my lawn and started talking to me like Gemini I would describe the aliens as intelligent.
What if it talked like GPT-2?
People are arguing it's human level intelligence...
I also like this: https://laurentiugabriel.github.io/token-town/
It shows internals of an LLM nicely, simplified manner.
I suspect like most you don't appreciate how terrifying statistical relationships become when you have truly vast data sets to train on... and also that we as humans aren't as shockingly unique as we think (compared to other humans I mean).
While being very capable, AI is missing something required for true intelligence and I struggle to explain exactly what it is I see missing.
It's not really "creativity" because much of that always was derivative in my opinion. And LLMs are (for some definition of the word) fairly creative as far as taking known elements and re-arranging them.
I think what is missing is sort of a world model building capability. As humans we see phenomenon and classify them informally and model "what would it look like if this were the cause of that?" type scenarios. We see qualities in phenomena and realize this applies to other things even though the things may be completely different. We run informal "thought experiments" sort of. This is hard to duplicate because a lot (most?) of it occurs outside of systems of symbols like math and language with fixed rules in my opinion.
Anyway yes, lots of human thinking is statistical and LLMs have that down pretty well but they are not "smart" I have concluded and it might be a very long time, if ever, until they are. That isn't to say they aren't very capable tools which they obviously are.
So, right of the bat, you are warning us that you are going to apply the " no true Sscottman" fallacy, and that we should brace ourselves.
Yes, models posses intelligence, but it is not a true one.
Then you claim that models do not posses world-building capabilities. But this is simply not true. Even ignoring the whole subgenre of scientific papers on exactly that subject, it is not that hard to build some hypothetical scenarios, big or small, and then witness the ease with which models do navigate those worlds.
Yes. And they are criticizing a model for not having a default mode network - as if that is some impossibility rather than just an artifact of the current iteration of the specific architectures we have built so far. Why do people paint with these broad brushes over relatively specific complaints?
LLMs are likely for machine intelligence something like drosophila are to biological intelligence - relatively early on the high dimensional spectrum of possibility. Though it stikes me that in a different way they're little alike - drosophila are relatively small and efficient.
How can human thinking be statistical when it is entirely based on ones lived experiences?
When people pretend to know what they are talking about - sure - but even that is not probabilistic - that is the person babbling together mush from their lived experiences.
Statistics has nothing to do with it - these are abstractions humans have invented to try and look at our surroundings objectively.
I'll restate because both objections (which apparently skim instead of read) are missing the important point. Yes LLMs can run "what ifs" scenarios and build models.
However LLMs deal entirely in symbols. 100%. Humans can "world build" aside from this and in fact are often at their best doing so.
Did the first humans to use fire and some form of a wheel even have the capability to talk about it? Think about that.
> However LLMs deal entirely in symbols.
They use tokens as input/output encoding. They do 99.9999% of processing in a high-dimensional latent space.
How exactly is this high dimensional latent space represented? Pixie dust and ethereal forces? Or floating point numbers? Where does it get the weights? Reddit?
Do you hold your experience of "dogs" (for instance) as floating point numbers? The fur, the fear, the love, the wet mouths, the sounds and colors?
Isn't thinking in images (visualizing non-verbally) also dealing entirely in symbols?
I don't think statistically driven prediction implies reasoning or intelligence.
Its a mirror to human intelligence. Regurgitating phrasing to match what someone who can reason put together, but it isn't any more intelligent than the reflection of you in the mirror is.
First you need to prove that human cognitive function is also not fundamentally driven by statistical processes.
It's possible that "statistically driven prediction" is all we are.
LLMs are pattern prediction systems with a large training data set. It is not surprising that they can predict patterns, particularly for a well structured field like mathematics that is also amenable to automated proof checking to help steer it.
"Pattern prediction" is a very broad stroke. What's something AI can never do that would astonish you if it did?
"AI" is a vague term that encompasses both traditional GOFAI, LLMs, and future tech so your question is meaningless. Non-human intelligence is possible, if that's what you're actually asking.
What part of my statement do you take issue with: that LLMs are pattern predictors (that's literally what the algorithm that runs it does) or that mathematics is rules-based and checkable and therefore amenable to automated pattern prediction?
Perform the same tasks as a human can, given only the amount of training data and energy that a typical human has access to.
> The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever.
I feel like that what a lot of people who say this don't seem to grasp, is that despite this flaw its still often capable of saying more interesting things than a lot of humans. Which says a lot about humans.
Idk something about a mirror maybe and the output reflecting the input?
> I feel like that what a lot of people who say this don't seem to grasp, is that despite this flaw its still often capable of saying more interesting things than a lot of humans.
So does the Google search bar, but I don't ascribe intelligence to it.
The Google search bar is not capable of generating original text though. LLMs definitely are - you can get pretty creative output easily.
If the Google search bar ran results through a thesaurus, it still wouldn't be intelligent.
String concatenation will generate "original text" by that definition.
Obligatory "yes, I know that's not what an LLM is", purely pointing out the metric.
The google search bar is surprisingly smart sometimes. What's your definition of intelligence that completely excludes most of what a computer does?
Happily, I don't need to define "intelligence" here, because it's squarely in "I know it when I see it" territory. It's notoriously hard to define.
I also don't ascribe intelligence to a pocket calculator.
I don't find it's particularly hard to define loosely, but then I don't think of it as a special property of humans other than it tends to be quite high in them. But we are obviously talking about different things and if you're not going to provide a definition then it's not really the basis for a productive conversation.
To address this similarly to my sibling reply, I don't have a definition of intelligence that provides value here.
And your loose definition isn't doing a lot of help either, beyond perhaps noting: that Google search bar _is_ similarly "intelligent" to an LLM? Which says what, a lot about search? A lot about modern LLMs?
Similar in kind, but not degree. An LLM is a lot more intelligent than the google search bar.
Do you ascribe intelligence to a gorilla? How about a goldfish?
Would you ask either of them to review a PR? Or a calculator to eat a banana? Or an LLM to calculate prime factors?
These aren't interesting questions. As much as any definition is in use here, we're not going to get much value talking about "intelligence" this way.
I mean I'm quite proud of some of my search queries in the same way I'm quite proud of some of the LLM output I get. I'm probably just very arrogant and enjoying myself via some LLM indirection.
Am I the only one that sometimes reads back particularly good emails they've written? I feel like its a similar thing :).
> often capable of saying more interesting things than a lot of humans. Which says a lot about humans.
Other humans aren't there to entertain you, the LLM is.
Most of these arguments are over some metaphysical definition of the word “intelligence”.
As per later era Wittgenstein, I prefer to ignore these engagements and focus more on the meaning-as-use approach.
What is the use of intelligence? What are the concrete outcomes of intelligence?
I strongly agree. Is a human in a vegetative state sentient? What about when they're asleep? What about someone with brain damage? What about someone with an IQ of 20?
Ray Kurzweil argues sentience is a philosophical question, and doesn't have much value as applied to science and technology. What will change the world is how this intelligence is applied. No one's going to care whether AGI is defined as sentient when it creates cheap fusion energy.
This seems like a complete waste of time given the more practical and more urgent need to clarify to everyone involved that current LLMs are not actually intelligent.
What an utterly unconvincing call to action.
You’re not offering a rebuttal, just making another metaphysical claim about “intelligence.”
You don’t even attempt to explain what practical distinction your use of the word is supposed to capture.
"The fundamental cause of the trouble is that in the modern world the stupid are cocksure while the intelligent are full of doubt."
— Bertrand Russell.
I'm not calling you to action, I'm explaining why I don't feel inclined to engage in philosophy and discuss "the concrete outcomes of intelligence" given a more pressing, pragmatic need.
It feel it's self-evident that we must fight the good fight of dissuading as many people as possible of the notion that LLMs as we have today, and likely forever after, are actually intelligent. Delaying this fight allows the current, stupid belief to the contrary to fester.
I don't think we'll win the majority of people over by debating the nuanced meaning of the word intelligence to a very precise degree.
I think we ought to do it by shaming them every time LLMs fail.
I think airstrike is joking. :)
I appreciate the charitable interpretation but I was not even joking this time! :)
Oh, in that case I agree strongly with William. I think the definition of intelligence a complete waste of time and the only real question is, can these tool solve problems for us? The answer is clearly yes, there are some problems they can.
The really interesting question is still a few years away when we ask if we humans have the right to turn these things on and off? ;)
> The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model — an algorithmic construct
Isn't this how human brains work? We're just large probabilistic neural inference machines. A network of neural weights guided by past training.
To be honest I think the debate about what "sentience" is is inconsequential navel gazing. There are many different kinds of intelligence - even within humans. What matters is how useful that intelligence is as applied to solving real world problems. Like it or not, LLMs produce intelligence which is very, very useful for 1.5B people and rapidly growing.
I think this ultimately boils down to the classical economic debate of marginal utility. There isn't an objective way to value a product or service. Each person decides for themselves what said product or service is worth based on their needs and preferences. Intelligence works the same way. We don't have the right to tell someone that their perception of the value of that intelligence is wrong. They alone determine that.
One thing we can all agree on, is that the capability of this intelligence is expanding rapidly. In a few short years, we went from Will Smith spaghetti hands to full length movies and strikingly realistic images. For 60 years, passing the Turing Test was considered Star Trek level science fiction. Last year GPT 4.5 passed the Turing Test. AI is already being used to convince people over audio that they are real, and very soon, this will occur over video.
I think people are being too dismissive of this intelligence. It doesn't need to be perfectly humanoid to be considered intelligent.
>Isn't this how human brains work? We're just large probabilistic neural inference machines. A network of neural weights guided by past training.
Artificial neural nets are merely cartoons of how people guess human brains work. They're likely much further from reality than, say, the fundamental laws of physics which can be experimentally verified or disproved.
We have actually little idea how human brains work.
Every age thinks they know how they work, and then every subsequent age laughs at the previous age’s rudimentary understanding.
We gave a Nobel prize to the psychiatrist for inventing a method to remove the frontal lobes of a brain through the nose in 1949. Lobotomies were performed through the 70s.
We are likely doing similar if not subtler but worse things today (just one more pill bro). We still have no idea what we are doing when it comes to the brain.
Yes and in actuality we dont need to know how humans brain's work to make progress.
Just like how airplanes don't flap their wings.
Is this progress in the room with us right now?
The state of population level mental health seems to be declining in most measurable categories.
As will be obvious when an alien intelligence makes itself known.
Why is being statistics/algorithms wrong? What's wrong with that? The "A" means artificial so none of this seems surprising or weird or bad.
The statistical nature means that what an AI produces is basically the average across all training data, making the generated text extremely bland and personality-less.
Don't discount the capability of representing human-like intelligence with simple constructs. Give enough parameters and advanced enough training you can without a doubt create real intelligence. We've seen some of this already with "j-space" where llms have started to exhibit reasoning before it ever reaches the output head.
AI discource is irrevocably fucked due to posts like these 2023-tier opinions.
"and does not possess any real intelligence or critical thought whatsoever."
unfortunately in most companies this is literally wrongthink and will get you shut down as being a scared luddite.
AI is just a good permutation/combination engine that tries to act smart with help of statistics. At best I only see AI as, 1. An autocomplete on steroid, 2. Good search/correlation engine
A decent correlation engine is still extraordinarily valuable for science, investing, prediction, etc. Plenty of human minds are strong in the same area.
Are we sure there is some objective, technical definition of what is intelligence and what is not?
Isn't it rather a subjective philosophical concept? What if human intelligence is also a statistical model, trained by evolution to make decisions that lead to offspring?
The one major difference I see between AI and people is the ability to learn and memorize. All memory/learning solutions that current AI architectures offer just feel like workarounds and simply don't work anywhere near as a person learning something new and remembering it.
I am firmly realistic on the overhyped nature of AI, but which discoveries in the last 100 years are not "a composition of solutions humans have developed and documented elsewhere" ?
Is that basically every new discovery? And under the strictest definition of novel and NOT falling into your composition of previous solutions what is that standard of proof to beat your criteria? Is a novel discovery not allowed to use English but must invent their own language? Must they invent their own math - these are hyperbole for illustration but I think its not far from that before you could just argue anything based off it is a composition of existing ideas
> The "something deeply wrong" part about AI, that even most technology enthusiasts evidently do not seem to grasp, is that it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever.
Not that I'm saying AI are like brains, but can you describe why brains, which are fundamentally slightly dodgy electrochemistry with frequent literal delusions of grander, are not "statistical"?
> No matter how much investors and tech companies want you to believe that they are on the verge of super intelligence, nothing I've seen to date can not easily be explained by "correlation engine", including the "novel" math solutions, all of which appear to just be "a composition of solutions humans have developed and documented elsewhere" upon deeper inspection.
Ditto, when do we humans do things exceeding the parameters of "correlation engine", especially if you consider compositing things either we or some other part of nature has developed and documented elsewhere to be insufficient?
I see what you did there with the —s!
Thank you for helping me keep my sanity.
> it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever
How could I seriously repeat that prayer, when it builds things I wouldn't be able to build and solves problems that I wouldn't be able to solve? I would have to assume that nothing I did in 25 years for money required any intelligence or critical thought whatsoever and I have higher IQ than 99% of the population. You might be comfortable with that but I'm more comfortable with ascribing at least some intelligence and critical thought to AI.
not having any body, continuous sensory input (except ChatGPT-live gets streaming audio), episodic memory, on-the-job learning (live weight updates), any live feedback loops (like moving a motor updates proprioperception or turning physically changes what a streaming camera sees), really hurts these models' abilities to perform tasks of the kind you're waiting for.
They are very good at instruction-following and you can teach it a new task that fits in its context and it'll learn it and do it. Go ahead, you can make up some new brand new ruleset or behavior and instruct it to follow it and it will. That's amazing.
But it won't be any better at its new behavior after an hour or ten hours or ten days. It doesn't have the kind of adaptation that we expect.
What it is able to do already is pretty amazing, but what it lacks is also a great hindrance to seeing its full capabilities. We just have to wait until research labs add these missing components.
Alright, take it easy. You typed a lot here but you're not actually saying much. LLMs produce useful outputs, their usefulness is just proportional to how well you know how to use them. Everything else is navel gazing.
> is that it is still fundamentally a statistical model — an algorithmic construct — and does not possess any real intelligence or critical thought whatsoever
What makes you so convinced that a algorithmic construct of neural nets cannot be "real intelligence or critical thought"?
It doesn't matter whether it theoretically could be. It matters whether current ones are.
I mean this in the kindest way possible, but you are wrong that the math solutions are that easily dismissed. And there are many more than are publicized. A specific math problem I wanted solved for 3 years did not get solved by any model until fable and, and I tried it on every model and know the literature surrounding it well.
Completely agree. AI is very impressive in many ways but there is something deeply wrong that is hard to put into words. The output is probable but never true, if that makes sense.
I think this is also the mechanism behind why AI generated videos and images are so captivating at first. I remember when Midjourney first launched and it was hours and hours of a brain-melting "Wooooooow". But once you get used to it and start to identify the patterns the brain quickly labels most AI-generated content as blank space.
If the image or text wasn't created by a human, then there was no intent behind the content, there is no message or novel information conveyed, and it reads as noise.
Yeah AI generated content hints that there is a whole world behind it, the way that an image pre-AI was a clue that there was a rich 3D space that corresponded to the image.
It seems our brains are adapting to that and recognizing "actually the signal behind this message is quite sparse" even when presented with rich imagery.
> But once you get used to it and start to identify the patterns the brain quickly labels most AI-generated content as blank space.
I guess the majority of people do low-effort generation that doesn't perturb a default style of a network enough, so it stays blatantly noticeable. The percentage of "super-recognizers" who notice almost all AI-generated images is around 1-2%. It could be that you are one of them, of course.
I think this is the confirmation bias trap a lot of people fall into; higher quality, hard to detect AI is already ubiquitous but because it's hard to detect people just don't clock it.
"I can accurately detect 100% of AI generated images that I recognise as being AI", if you will.
This is true, and doesn't even get into low-effort generation (and the 'perturbing') slightly disguised and then no one sees it at all.
> If the image or text wasn't created by a human, then there was no intent behind the content, there is no message or novel information conveyed, and it reads as noise.
If I were to push you a bit on this, when is it not true?
Let's not like at AI specifically, but can you think of other examples? Like for me, I think of: the creation of earth itself, or stars, or even DNA.
I think you are adjacent to the real story here, but missing it. AI text contains information, certainly. Frontier chatbots are very good at creating acceptable and mostly accurate answers to our questions on just about any topic. It’s an astonishing achievement.
But you are sensing correctly that there’s something missing. It’s the meaning and the speaker. Communication is an exchange between speaker and listener. The speaker has a meaning in mind, and wants to create that same meaning in the mind of the listener. Therein the problem.
There is a listener, sure. But no speaker. No meaning. There is information, but how can this be communication? Nothing is talking. Or at best, we are just talking to ourselves, our own words back at us through the funhouse mirror.
When your mind looks at AI text, you know you can safely ignore it. No one wrote this. No one cares if you read it. You can delete it and nothing of value will be lost. It might contain the information you need, or a bunch of gibberish. There’s no one’s reputation on the line if it’s gibberish.
Yes AI content, even when correct and insightful, has a feeling of being disposable. Maybe it's because there is no person behind it.
I prefer to think of this in terms of Umberto Eco's opera aperta (open work): if any text is a collaboration between author and reader/recipient, here, all the burden of meaning is left to the recipient. There's simply no meaning on the side of the "author", it's just a statistical extraction.
(There's also the problem of words/signs (just) referring to other words and/or cultural entities. There is no world nexus in this, therefore also nothing we conventionally refer to as meaning. On the other hand, it's utterly dogmatic, as all it refers to is the most probable construct, as a reference to references that are just another utterance, but supposedly a dominant one.)
For me it's youtube videos. As soon as I hear the AI tells in the script, even when it's clearly read by a real person I immediately look for a different video.
At least for the content I watch for entertainment, it may be different if I am looking for a specific answer for something where I would otherwise just ask an AI anyway.
Yeah, I have the same problem. There's a good quote example of this:
> There’s a growing scissor between people who are happy to read AI and those who violently bounce off from it.
> People adapt in different ways — and some people absolutely cannot look at it. That cognitive split creates a surprisingly powerful opportunity: you can write something that, technically, sits right there on the page, yet an entire sub-population will be incapable of staying with it long enough to actually read it. You can hide entire sub-structures in plain sight. It’s not avoidance — it’s adaptive obfuscation.
> The paragraph before this one was the only thing generated in this essay and if you just skipped over it I highly recommend reading and really understanding what it’s saying.
It's quite effective. I think this kind of text functions like the chumboxes you see at the bottom. Taboola and so on. Just mental ad-block takes over.
You are re-compressing information that is in-effect meaningless because it's all decompression artifacts.
The AI had a nugget of data and decompressed that into a flood of text.
The exhausting thing is that we're then trying to re-compress that or derive the original intent and meaning from noisy decompression.
It's like un-zipping a zip file into a probability space of what could have been in the zip -- and then having to find the actual files worth reading.
For some research I looked up some very old Reddit threads a couple of days ago.
And, Oh my god, you can actually see how this style of writing influenced AI writing today, I constantly had to remind myself: "this was posted before ChatGPT released".
The reddit influence is especially true for "storytelling" writing.
I experienced the same lately. Even dug some of my old posts where I put in the effort and formatted them using reddit's markdown. Wouldn't dare it today
When I read AI-generated prose that is aimed at the general public, I have the exact same feeling.
But when I ask Codex a technical question about coding, I don't get it at all. Codex replies to me in a very direct, technical manner, similar to the way I speak.
When I ask ChatGPT to be concise and technical, I get the same effect.
I think it's because prose aimed at the general public has to be very attention-baity --like the textual equivalent of a Mr. Beast video--, not because AI is incapable of writing like a human.
I use Claude and I find that it speaks in a very obfuscated manner when explaining things. It seems to make up jargon as it goes on top of spending a lot of tokens dancing around a point. I often find myself having to ask it to rephrase things, or speak directly about mechanism or consequence, in order to understand the point.
Using Claude for any kind of technical writing makes me feel like it was trained on snarky Huffington Post articles written by a 23 year old mixed media arts graduate and then was told to intentionally obfuscate the most important elements of any text by extensively rambling about what was not done and for what reason.
GPT is less bad for this, which is why I've mostly shifted to using it.
Not to accuse them of doing this, but AI vendors have an incentive to generate verbose responses, given that you pay per token
Beware it's not baiting you in the exact same manner but different particulars.
I once heard on a podcast about comics that an artist ran into someone who couldn't read comics. Not that they didn't like them, but they claimed that they just didn't understand how they worked. When the artist explained 'well each panel shows an illustration of the events in the story', the other person was confused - "each panel is connected to the previous?" they asked. They just couldn't wrap their head around the fact that each of the panels was connected in some way - their brain just refused to interperate them as anything other than 4 separate completely unrelated panels.
When I see AI animated videos, that's how my brain feels. It's this strange brain-fog that I just cannot connect together the sequences of images being shown into some sort of chain of events. My brain just refuses to see them as anything other than a disconnected series of 2-3 second videos, even if the same character(ish) appears in them all. It's very strange.
Do you have much exposure to pre-AI corporate memos, mission statements, marketing plans, or white papers? Because they were mostly written in that style. Full of buzzwords, cliche similes, platitudes, jargon and stock phrases.
The thing is, people writing them had a style. Every company has its own style, or feeling for these kinds of texts. Also for the initiated, these buzzword-filled blocks of text provided some between the lines information; sometimes big, sometimes small.
AI generated text doesn't have this. Every model has its bias towards a certain style, an overly agreeable tone, some exaggeration to make the user important and smart, but the text has none of the information crumb these pre-AI texts contained.
Even when you use tools like Grammarly and allow it to "Impact-MAXX" your text, the resulting text is a bland wall of letters, carrying none of your voice or style, less elegant than a corporate text and emptier than space.
It's beyond bland. It's tasteless.
AI tries to make the prose "interesting". I don't want to read interesting prose. I want to read interesting ideas.
The prose is not only interesting, also glorious. Gloriously grandiose, monumentally empty at the same time.
It's like a hook of a pop song. Interesting to listen, but entirely empty.
I also read right through those. 90% of most companies' webpages is drivel. Same with 90% of a job listing's text. I hardly understand how people get any meaningful information from corporate websites, they're all just "enhancing your business outcomes by incorporating technical excellence with synergistic AI" or something like that.
There's somehow less information than if they just asked claude to make something up without any context.
> There's some psychological mechanism by which my brain immediately recognizes AI generated text and just short-circuits to "there is no information here".
The roots of llm math in part lie in compressing natural language such that there's only information there, and then running the reverse to create way more text without new information in a somewhat precise theoretical sense.
Some more information: https://youtu.be/l6DKRf-fAAM
I love 3b1b and I love that video, but that also isn't exactly what is being said. In particular llm inference does add information (in the meaning in this context) because the output distribution is sampled randomly.
Back and forths with the LLM are very useful when the person inquiring the LLM is invested in the conversation and wants to uncover the ground truth. Ultimately, some good result can come out of it because the person has a goal and LLM helps them reach that goal by uncovering the layers of knowledge which the person might not have.
When the AI-generated content is presented to a person without any prior investment, it just looks incoherent. An especially great example are these Claude-generated explainer-type pages, which look really nice, even interactive, the information from the first sight looks really well presented. But somehow it all just doesn't make sense to a human. And I think it's because humans are processing information linearly and building an internal story about the information. One could argue that LLM's also consume information linearly but the way this information is processed is a kind of all-at-once approach.
Just some speculation on my part but I have been trying to cope with this way information is presented because I am currently working at a place which is heavily documented by AI. And the only way for me to properly understand the documentation is by inquiring AI to help me.
> Something is deeply wrong with AI generated output, and I say this as someone who is typically very impressed by AI.
It is because GenAI output has no thought behind it, as you identified in your previous paragraph:
> And when I force myself to read AI-generated text I realize I'm making my brain do creative work to impart meaning to the words. It is exhausting because my brain is literally trying to do a just-in-time rewrite of the text into something valuable.
You are searching for meaning in something which was not created to convey meaning. The text was, instead, the result of an extremely clever statistically based algorithm.
Not contemplation. Not thought.
Oh my god thank you. I have been trying to put into words what is so draining about pair programming with claude, and “doing creative work to impart meaning to the words” is exactly it.
I kind of wonder if our ability to skim has been stymied.
blah blah blah
- blah blah nugget blah blah
- blah blah blah wrong blah blah nonsense
- blah blah blah obvious blah blah
- blah blah blah off-base
blah blah blah
It is that we HAVE to skim because the text is so cheap, and it wears us out.
I think we as professional documentation-readers already skim most content (speaking for myself, I realized I was googling and skimming for answers 20 years ago instead of reading documentation end to end), but AI generated content has the same problem as marketing speak in that it's a lot of fluff.
It's understandable people don't read but feed stuff into their own AI again to bring it up to their standards or have it get to the succinct point.
People should notice that it is constantly inventing plausible jargon, some of which may or may not have been used in some specific context.
It gets worse with language mixing, but I can't help from finding it funny at times, unless it bites me.
Yes, I've had both ChatGPT and Perplexity return English answers with Hindi words sprinkled in (for totally unrelated queries).
For example, I asked ChatGPT to summarize a long news story and it substituted the Hindi equivalent हत्या for the word "murder", as if ChatGPT was trying to work around alignment training or keyword block lists that discourage it from using the word "murder".
Yeah that's a very good example, because it also demonstrates the "alignment issue", assuming ChatGPT wants to avoid confirming accusations of murder, or simply using the word without strong evidence.
So kinda charitable :)
I was recently wondering for a minute, shame on me, what "the stand of the deployment" means, because in the given context, it was almost halfway meaningful to consider the AI thinking that the deployment "has a stand" on something, when compared to the development environment.
Jargon is even worse though, and I've not yet verifies whether it gets reinforced by language mixing.
"Decider-verifyer resolution" was kind of neat, however, it wasn't some sophisticated machine, it was the verification loop I agreed on with the AI (mix of tools usage and manual steps).
Just the other day I was using text-to-speech with Gemini, and for some reason, it transcribed my full query in Hindi (in the middle of an English conversation), and naturally the LLM responded with Hindi as well.
I don't know exactly what I said, but after translating it back, it appears to have attempted a phonetic transcription of my words (rather than translating my actual question).
I wonder if this is because of all those YouTube videos with the title, description, and language set to English and the content in (presumably) Hindi. I run across these a lot when looking up obscure topics.
Good to know that at least Gemini hasn't forgotten about its true roots :)
Are you sure you are not doing the same thing with other texts?
I started to skim a lot more text due to me having read a lot. Like in news article, i stoped reading the first paragraph because it repeats just what it was already written in the short subtext. Then there is the second paragarph which is used to have some historical view or whatever it is.
I am very good at skimming over text. Human-written text I can usually glean the gist from very quickly, and get to choose how much I want to glean from it: The closer I look, the more I find.
With AI-written text, it's almost the opposite: the closer I look, the less I find. It is so information-sparse.
I started skimming reports im required to produce quarterly snd annually. I designed them to provide novel information at start and end so I can update them easily.
The problem I encounter is both my memory is degrading, but since these reports are largely duplicative, knowing which version im remembering is technically impossible since theres so much overlap. The overlap is tge same problem as context poisoning.
Id been doing this for over a decade when i started working with a new engineer with a few years of experience and younger. I tried to explain how i set these docs up so they can be skimmed and you can update the specific facts needed. They exclaimed they would never skim and rewrite it all. There was zero way to explain how exhausting that will become as they age.
So theres certain a tension about how people and AI will generate documents.
Let them discover it themselves.
Interesting anecdote!
It's a machine that generates plausible letter sequences that we all pretend have profound meaning.
Sometimes they happen to be correct, but you have to read them in excruciating detail to know that.
The junior engineers at my job have a terrible problem of writing AI "proposals" to problems. The proposals are all extremely detailed and verbose to a thought-terminating extent. It takes a lot of effort and self-control to parse out the actual "ideas".
I think of the Dwight Eisenhower quote: "Plans are useless. Planning is indispensable."
The process of thinking through a system and communicating your design to other humans is a core part of software engineering. You want to build the right abstractions and communicate the right level of detail. Delegating all that thought to an LLM means your proposal isn't clear to the target audience, and it's not helping the author to understand the problem.
Isn't that mostly hyperbole?
When I ask AI to research technical information about X and (include sources) - I get mostly solid information as response.
But poems or interesting fluff blog entries by LLM's? Not something I look for.
What disturbs me is all the "pretending to be human" all the personalizing language - that is clearly fake and I would much rather have a neutral robot language as response.
Another good example of AI making my brain do more work is when you ask it to compare two things:
“Compare a car and a bicycle”
The answer is invariably something like:
Instead of “bikes are useful for short trips if the weather is ok and you like getting exercise, whereas a car is usually better for longer trips, bad weather, or multiple people”I wonder how the autoregression works on that. You'd think "seats: a bicycle has 1" would be an easier autoregressive completion than "seats: 1 (bicycle)" since it works left to right adding context.
Same. Also when I see a spec for example or some summary I always have the impression it doesn't get to the point. Like, the core ideas are in there but also somehow lost and I have to work them out again which makes me wonder if the person generating it understood what is going on or if it would just have been fast to just write it by hand (you still use LLMs for research and such).
yes, but now I’m also experiencing that for human-written text
> Something is deeply wrong with AI generated output
For me, it is the endless maximalism and hyperbole. Almost if the output was driven through a radio-mix compressor - too loud for the reader/listener to be able to pickup any dynamics.
It's the YouTube clickbait algorithm applied to text. Maximize interaction, at all costs.
> just short-circuits to "there is no information here"
That is my experience with the way the models write by default, often even when instructed not to do that. With enough effort you can get even them to slightly unslop the writing so it doesn't read like some LinkedIn/Buzzfeed brainrot, but the problem is that it's not trivial to do and most people won't do it, so the default is indeed horrible.
Once you see past the illusion I think there’s no going back. AI writing style is just dogshit. This hype wave is based on the belief that we’re inching closer to AGI but seems to me we just increasingly struggle to define intelligence. LLMs seem smart because they can pump out thousands of LOC quickly, and enthral you with fancy words and bullet points. I don’t fall for the intelligence illusion anymore.
I'm not sure we need to declare AGI around the corner nor declare it all dogshit. I think that's part of what's so dissatisfying about it; it strikes at such extremes of both awesome and awful.
> There's some psychological mechanism by which my brain immediately recognizes AI generated text and just short-circuits to "there is no information here".
I think you need to self-correct here, because otherwise you'll be ineffective in an information setting, where I expect AI-generated resources will not only be the norm, they will absolutely swamp the environment.
AI-generated resources swamping the information environment only makes it more important to have the mental mechanisms for quickly filtering out their non-information.
Yeah I don't think the solution to a flood of useless information is to try and digest more of it.
I'm not saying digest it, I'm saying be able to scan it/skim it/move on, but ignoring it won't help.
Perhaps we'll all become metaphorical pandas, spending 14 hours a day ingesting nutrient-poor bamboo. (And producing a proportionate amount of excrement ourselves.)
I hope not.
Do you have the same feeling when you chat with AI?
It's that you know it's a waste of time. If I sent you emails that were full of nonsense every day, you'd start tuning me out too.
It's like if on any website you went to you saw a lot of posts written by the same guy over and over again. Even if he used different names, you'd start to recognize him eventually because of his style. Seeing as he doesn't say a lot of valuable stuff, you'd also learn to skip whatever he says.
I do worry that it's just survivorship bias and we're also consuming higher-quality AI output that's indistinguishable from human writing, but we focus on the raw, unedited, low-effort AI slop and think that we're good at recognizing AI text. Even if we really are at the moment, it might not be long until AI companies figure it out. I'm not sure why they haven't yet, given how many books they've burned for this already. Maybe it's just more efficient for the model to stick to a single way of writing, I don't know. But when that point comes, we'll be back to the usual way of reading and interpreting text because there would be no way to tell what produced it.
It reads like the white papers companies publish on their websites to build legitimacy. Or anything from those IBM / SAP / Deloitte / etc consultants who write technical papers despite having little to know understanding of the technology.
That's why the business and government people love it, they spend their entire careers reading this nonsense.
I feel you. Everytime I need to read a long text I just assume it's AI nowadays and I hate it because most of the time I'm right.
I hate how AI writes. How much numb filler bullshit is in content I need to go through for my job.
we are working on it, the thousands of gig workers tuning frontier models
Didn't you have this mechanism before the AI? The internet is a cesspit filled with empty words and commercial speech for some time now.
Yep. It's like it's painful to read for me. It's because the next-token predictor is just mashing (mostly) grammatically-correct and plausible sentences together, without any real intention or meaning. So everything sounds plausible, but almost entirely void of meaning.
I've got a 3 step instruction to compress Ai text into useful info.
1. Ask it to write according to the Google Developer Documentation guidelines. Gets rid of fluff, less emotional statements, no it's not x it's why.
2. Tell it you have extreme ADHD and need everything condensed as much as possible. You can always ask for expansion on an answer later.
3. Bullet points whenever possible.
>and just short-circuits to "there is no information here"
I feel the same way when I read a "press release" or anything written by marketing. Even the newspaper will only have 2-3 sentences of interesting information spread out over 4 paragraphs.
This makes me think of the paper-to-media pipeline; scientific papers are high information density. Its press release summarizes the finding. Then the popular science and social media posts come that oversimplify and embellish the findings.
So from "this table of stellar luminocity observations shows x y and z" to computer renders of green/blue planets with captions of "LIFE FOUND IN SPAAAACE!".
Before AI slop, there was corporate slop. It served the same purpose and was generated in largely the same way. But it was only deployed where it was worth the cost of creation.
> Something is deeply wrong with AI generated output
It works just fine for me.
You are absolutely right.
haha this made me laugh
That's not funny, it's serious!
This is just a weird feeling that I've been coming closer to articulating lately, but I only think that you can get forward reasoning from what is basically word association; there's no mechanism for unwinding it because it has no real memory. By "it" I mean word association itself, not any context window. It predicts what could be in a position, and ignores what wasn't in a position.
People don't do that. People are constantly engaging with paths not chosen. Right after I choose to write one thing, I'm immediately engaging with what I chose not to write there - I'm explaining why I didn't write it, I'm realizing that my choice may seem unusual so I'm trying to make it memorable, I'm focusing on the distinctions between what I wrote and what I didn't.
LLMs don't currently do that. LLMs just ape a structure. When the structure resembles the sort of timid, clarifying fussing I just described, the LLMs just drift randomly because what they didn't say wasn't in the context.
I also think that's why they have such a serious problem backtracking. They're not taking into account the already eliminated possibilities. Often the thing that was so unlikely that you weren't going to waste time on it is the answer, and things you discover while going down an ultimately wrong (but initially far more promising) path remind you of the path not taken.
They're simply assembling a thing that resembles a valid argument, and happen to make sound choices because the plurality of input happened to contain sound choices. This is usually a very good bet because there are so many more ways to be wrong than to be right. But it doesn't account for attractive (common) wrong choices. You need a way to back out of those.
Exactly, AI-generated text reads so smoothly, that the same short-circuit shifts my attention away from deep focus and onto scanning of the text, looking ahead to get the gist of it. Forcing myself to read the text fully feels almost painful. It's like reading a terms-of-service or any boilerplate document.
> my brain immediately recognizes AI generated text
I bet it does. I bet it also recognizes some human text as AI text, and doesn't detect other AI text.
I am not claiming to have a perfect AI classifier. That is an unnecessary claim that distracts from the broader point.
Show me AI text that manages to climb out of the uncanny valley, and I'll show you AI text that's been edited by a human.
https://github.com/blader/humanizer
The problem is the well's been poisoned just by the fact that I know this is AI trying to hide AI, so I'm already poised to look at the examples and declare "aha! this is obviously AI!" Moreover, it's not single sentences or phrases that make AI text stick out (though obviously those are the biggest tells), it's the text taken as a whole. When you read the full output example in that repo, it seems obvious to me that it's AI (though again, it could be the poisoned well). This is the uncanny valley I was talking about; something is just off about it.
I agree that it feels off and I wonder what I would have thought if I'd seen the "after" example without knowing it's AI output put through a humanizer. Would I think much about the weird use of the word "honest"? About "that's the Lisbon I kept thinking about, not the castle"? Or how the story feels very impersonal somehow, with the author just mentioning their calves and legs sometimes as the only way of convincing the reader of their humanity?
Also, the 'before' segment didn't contain any mention of custard tarts, football, crowded trams, mixed feelings, etc. The original had a very positive travel agency type of tone, which was replaced with a lot of very odd sounding, imperative phrases that sound like engineering-speak. ('earn the fuss' 'build trips around pastry') I'm not convinced that this thing is actually meeting its design goal of not hallucinating shit.
The readme feels AI generated
My son is currently learning Romanian and I was trying to help him with verbs. I don’t know Romanian but recalled when learning a foreign language for the first time it really helped me to break down how a verb form or tense worked in English, then learn the equivalent in the new language. So I wanted to make some charts and pages that he could use as learning resources.
I used Claude to help. I don’t know how to quite describe it, but because the text was polished and well constructed my brain was giving me the the signal “if you aren’t getting this it’s because you’re not focusing” so I’d read it again and then again and it still was not landing. It sorta felt like when you read something technical or heavy when very tired - you are reading but not processing.
Only after wrestling with this for a few days did I realize that it wasn’t me. As I started going through, sentence by sentence, forcing it to re-write things to be more clear the concepts became easy to understand.
I wish there was a name for this situation. It’s almost like a pseudo-language where it has the correct form and presentation but is missing critical components.
The more complex the topic, the more I sense this.
This is a really good description of the problem. I’ve been trying to use Claude to get familiar with the mechanics of a new codebase, and there have been so many moments where I’ve stopped after reading the same paragraph five times in a row and thought “Am I tired? Or stupid? Or is this codebase just wildly more complex than anything I’ve seen before?” before realizing that it’s just taken English and smushed it around like a ball of clay into some abstract sculpture that kind of evokes something from real life.
I think part of it might be an innate feature of LLMs, but Claude seems extra prone to it lately. I ran the same query about the same codebase with Codex, and it gave me an answer that was about 1/4 the length and made me realize that it really wasn’t all that complex.
If nothing else, it’s good training for my own writing. I’ve been working on making myself be more straightforward and concise, and Claude’s writing is a good example of how cleaner prose is a functional choice, not just a stylistic one.
It's a Claude problem, chatgpt is about x10 better and it's quite rare that I'm getting such foggy vague texts that I have to reread to understand. Chatgpt used to have the same problem in the early days but even then it was less pronounced than how it is in Claude today. Fable is a bit better but not as good as Chatgpt.
Ironically, I've been already active on codex after trying it for the past 1.5 months, from Claude, and I find it to be similarly complex and confusing.
I think they all have the similar styles and tells. If I were to go to Claude, and use it now it would probably be clear for a little before reverting.
And I don't know why it feels to me like the language "drop off" happens after some time with the system. It makes me wonder if my account are getting silently degraded or sent to lower intelligence/lower priority queues after being a member for a while.
> but is missing critical components.
The human spirit. When you read a real person's thoughts you can often intuit the thought processes that led them to write it which aids understanding. Or at least have a general idea of "where they're coming from". But an AI is missing that. It just knows everything, without a "thought process". Instead of a flawed 3d person, we get a nice 2d picture instead.
Feels like the way a video game will render the outside of a wall or solid surface, but you can run into it and warp partly through and there's nothing internal to it at all.
Like backface culling--as soon as you walk to see it from a different angle, the depth vanishes.
> I wish there was a name for this situation. It’s almost like a pseudo-language where it has the correct form and presentation but is missing critical components.
The best I’ve heard of this is peeling the onion. The first pass is always very high-level and you have to make it go deeper. That can be done manually with follow-on prompts but I like using subagents, each with a different angle on the problem.
> I wish there was a name for this situation. It’s almost like a pseudo-language where it has the correct form and presentation but is missing critical components.
Bullshit?
I understand the sentiment, but not quite what I’m thinking. It’s not that the AI is necessarily wrong, it’s more like, it replaces clarity with rich but unhelpful text. Maybe like candy. Full of flavor, texture and color but lacking the nutrients.
The way I like to describe it is that it feels slippery, like it's been polished down to oblivion and my eyes slide right off of it. There's no place to find a mental foothold, and if you zoom in there aren't any details.
But I also like your candy analogy because I think it's spot-on for how LLM text superficially looks informational/nutritious, even though it's actually just junk.
In German we call this "Hegelei". Using a lot of fancy words and complex sentences to obscure how little content they have.
No no no. Harry Frankfurt wrote at length the difference between liars (who care about the truth, and twist it) and bullshitters (who don’t care about the truth, but just the way they’re received).
This is some third category of untruth. Almost more sinister than the other two altogether.
I think, they don’t care about truth at all, just whether the reviewer rates it highly. Otherwise, it would be deleted after that round of training. It values test-taking ability over critical thinking.
My two favourite words for this are “conditioned” and “catechized” where the latter is a bit more on the nose but way more obscure.
"A third category of untruth" is a great expression, thank you.
A lot of it is hedging and using lots words to avoid saying something that isn't true. That way, it looks a lot like text that contains meaningful information, even though it doesn't. The speaker can pass the scrutiny of an informed audience, because they are able to substitute their knowledge into the words, as if by pareidolia. It's the sort of 'diplomatic' language you see from politicians, lawyers, students writing essays, and any other moderately intelligent person who is put in a position where they will face consequences for not answering. Behind all the fluff, there is a very loud voice yelling "I DON'T KNOW."
> It’s almost like a pseudo-language where it has the correct form and presentation but is missing critical components.
Slop. The word is slop. Has been for years now. I mean, is this not exactly what we've all been talking about the whole time?
I mean, yes - but somehow this feels like a different category. At least to me. I view slop as the classic garbage like:
[Thing] isn’t just [X]—it’s [more dramatic Y]. And [short validating statement].
I can see and smell this type of slop from a mile away. What I’m referring to is in the same family as slop but somehow different - it fools my brain by putting on the presentation of credibility and thus it is even worse. I can skip right over classic slop without much effort. This kind of text tricks me into laboring over it before I realize it’s hollow.
So in that way, it’s worse than slop.
At least in my mind, lack of substance was always the defining feature of slop. Stuff like "it's not just x, it's y" would be reasonable (if grating) tics from a human writer. Whereas humans are also capable of clever-sounding but substance-free writing, and it still sucks. AI just produced enough slop that we were forced to give it a name.
Technically, it's still slop. The first example is just slop that you recognise as slop. For reference, see gell-mann amnesia.
> I wish there was a name for this situation
Bad writing. The name is simply: bad writing.
Seems to me that Anthropic has done too much training on Hegel and Adorno books.
Uncanny valley.
I've been stuck in loop all week where a coworker asked me to review some methodology documents that are clearly generated by Claude (which is fine in this case) but every time I sit down my brain can't get through 2 sentences without feeling like I'm reading nothing. It actually feels like the content, which isn't empty, has no meaning. So I go on to another assignment and come back later and the same thing happens. And now I have anxiety about opening it so it has compounded in difficulty haha. Glad other people are experiencing this.
I had this the other day with a colleague who is Chinese. Their command of English isn't the greatest, but I never really had an issue understanding them before (in fact, despite their poor English it has literally never been an issue!), but now I just get these walls of text in typical LLM English that have become absurdly difficult to parse. It's like the uncanny valley of text, if you look at it as a whole it seems like it's a proper sentence. The words are all real words, they even make sense in the order the AI has put them in, but as a collective it's like my brain cannot connect more than a sentence or two together.
I thought it was maybe me just becoming lazier with reading considering how much AI-generated text I'm subjected to against my will, but I picked up Blood Meridian the other day and have unironically had less trouble parsing that novel than I have the majority of LLM-authored text I've seen.
I have basically polluted my context and agent files, agent.md, claude.md, gemini.md with really strong guidance about using plain language. I even include small, mini design specifications, examples of plain engineering language, only using technical language and invented terms when they actually make sense (almost never). EVERY TURN.
I have a pretty large set of prompts that go into any software engineering, and I force every single agent to use an ephemeral style stack of prompt management. So, every turn it goes to the top of the stack and it is the very last thing they see in terms of all of my prompts and instructions and agent files. And then it gets taken out of the conversation so that it doesn't get sent to the agent the next turn (no context bloat). It has restored so much sanity.
I tried the caveman add-ons, and I felt like I was losing IQ points because I spend a lot of time reading agent output, and when they start talking like cavemen, I start thinking like cavemen. That was not good for my mental health. So, I try and make the agent talk like me and think like me. And it works, mostly. And my observation is that maybe I'm not the most efficient agentic thought process, but my sanity is retained.
All of that is to say that if something is reading like that to you, just have the agent rewrite it and read it in a rewritten tone because it's probably bad as it stands and your colleague did not put enough effort in it. It is /not/ good and you should not accept it as a default. We have to hold the line on stuff like this and maintain some semblence of normal human engineering standards that existed before AI. They are not making us better. They are making is lazy and dumber.
Opus 5 and other agents in the latest rounds of tuning have gotten ridiculously bad in terms of how they feel to interact with with all the invented language and localized nomenclature. It is an obvious bias that big words and technical talk looks good to the bottom of the bell curve, but when you actually try and understand it, it's horrible. So people say, "Yeah, that looks great," in all the RLHF rounds, and they run with it because they think it looks good, but it doesn't. It's terrible.
Hold the line. It isn't you. And it isn't a good methodology document.
"nd then it gets taken out of the conversation so that it doesn't get sent to the agent the next turn (no context bloat). "
Uh - dude - this means you're paying 10x in token costs because there's no caching.
If you 're-write token history' then you can't cache tokens.
It means for any reasonably long conversation, the llm has to reprocess the entire history as preflow on every prompt.
Are you sure you're really doing what you say you're dong, and how is it not blowing up your budget?
Nope, you gotta understand. This is the move. It is appended at the end of every single message. It is never saved in the conversation that get sent back for inference. So you send it. But when you go back for inference, it’s at the top of the stack so all of the cashing works you’re not pre-filling every time or anything like that. It burns plus N tokens, where N is my prompt stack. It is not really that expensive. I have measured it to within an inch of it its life. Think of it this way every bit of the prefix and the conversation stays exactly the same you’re only adding to the very end of the conversation. So after the first turn, it is basically always cashing within the KV cache for a given context. Sorry I am using voice dictation. My hands are tired this week. Basically you sculpt the conversation history to ensure prefix caching
It should burn N + len(answer), because you have to re-cache the whole answer without the prompt stack.
Perhaps more persnickety, it pushes the LLM out of distribution - if it’s unnatural for it to write in plain language without the prompt stack, your prefix will be an unnatural conversation which can reduce intelligence in hard to measure ways, especially over long conversations.
Not saying don’t do it, clarity is perhaps worth the intelligence hit, but it’s not going to be a free lunch.
Yes, maybe. Evidence around caveman shows this isn’t a big deal for token consumption (forcing it to respond in a way it was not tuned) and I don’t see a big difference either way. And maybe it increases intelligence in hard to measure ways. Lots of parameters in these models.
I feel I fight them less with this setup. They get so lost in their own invented bullshit they stop being useful pretty often without it. So I would take bets on that :)
Yes - with arbitrary prefix caching that might work. That's tricky though, not everyone is going to provide that.
Did you have to build your own harness for this? Or hack Claude Code or something?
I don’t use their CLIs. OMP/OpenCode. If you are stuck on them, you can do it in a proxy layer.
Check out the awesome [Claudish-to-English plugin](https://github.com/gvzdv/claudish-to-english), you could lift their prompt from https://github.com/gvzdv/claudish-to-english/blob/main/rewri...
thanks for this, I'll give it ago. I've tried a few different concision prompts but haven't had great results
I don't want to read my own Claude outputs, much less someone else's.
If I had to, I'd process it into a short summary and/or ask an agent questions I have about the methodology.
I would then give my feedback. If they ask for details, I can have my agent update the document directly too.
I would not treat an AI generated artifact, be it documents or code, as something that a human should process fully manually.
Agreed, what I call "have your AI talk to my AI" [1]
I think that's where the future is going. It presents interesting challenges, but also some opportunities to make life better for everyone.
[1*] https://foundrs.com/have_your_ai_talk_to_my_ai.html
(*) I dislike the HN trend of starting citations at 0, I find it snobbish.
Yeah I think a bad prompt can create an output which is meaningless. I had such experience reviewing a software architecture diagram generated by Claude.
Worse, I noticed that people in an office environment themselves have adopted a more speculative, communication style.
In the past, people remembered what was said and would draw attention to discrepancies. I could trust what people said.
Nowadays it's like; someone can say one thing one day and the opposite the next day (through convoluted language) and nobody bats an eyelash. Or sometimes someone will agree with me but then what they say immediately after reveals that they didn't understand the essence of my point at all. I didn't notice these things 5 years ago.
I guess this is what AI researchers refer to as 'model collapse' - it seems to affect people too though...
I've noticed this too. Because people are generating work so quickly , and maybe not understanding it entirely, then they hit a bug, pivot, and pretend it was the plan the whole time. Or drop something they were very excited about 3 days ago and never speak of it again.
There is a lot more hustling and bravado.
It feels like people don't value knowledge as they used to.
It's really hard to avoid mistakes when everyone is subtly covering them up. It feels like a lack of care and I find it demotivating.
I think because engineers are afraid for their job, they are under more pressure to talk a big game. Also under more pressure to deliver short term visible results. Bad combo.
Holy shit, someone putting it into words
I'm worried we will all be talking to our Claude Therapists soon about our Claude induced learning disabilities lol
This is actually funny because I've been doing some RAG experiments and I've basically trained up a RAG on several gigabytes of philosophy and neuroscience. And when I want to go metacognitive about thought processes, I actually really enjoy exploring different areas of philosophy, more life philosophy, but how it intersects with psychology and neuroscience and observing the observer, or the observer is the observed, as they say.
And when you ground it with real data, it's actually extremely useful. It's not exactly like Claude Therapist, but it's sort of the teach me about philosophy, but actually grounded and not vied. I have a lot of really strict prompts and grounding and agentic guidelines for this particular agent flow and harness that I've built.
And it's just a few weekends of vibing and feeding it basically all of Wikipedia and several gigabytes of papers and stuff, but it actually leads to interesting discussion. So I just have my personal philosophy bot and it's pretty fun.
One of the modalities I built is having two agents assume a famous persona. And then they take a thing, like grief or some thing that I experienced during the week, and they assume the role of the two different philosophers, and I just have them go back and forth 30, 40, 50 turns. And it's actually quite interesting, and it really moderates their language and tonality and behavior. They really get into the roles when you have the right prompting and grounding. Sometimes they get a little off the rails, but it leads to genuinely interesting areas to explore, and then I'll actually go read source material and things like that. I don't know, that's how I do therapy these days, but I never actually did therapy, so I just think a lot, now with agents finding interesting stuff to think about too!
I also find it impossible to parse half the comments that Claude tries to sneak into our pull requests. I’ve never had an issue understanding code comments written by humans like this before. The structure of the information is like a waterfall that leaves me unable to swim to the surface and grab the air of comprehension.
So, “Please write a one-liner comment manually to replace these 5 lines of AI generated comment” is a common refrain in my PR reviews to colleagues.
We're using grok and while I can usually understand the comments they're always at least 2x as verbose as they need to be. It loves explaining everything in two different ways, putting one of the explanations in parenthesis.
It is getting to the point where you are better off getting an LLM to describe the PR changes in your preferred style (make it very short, make a table describing API changes, list renames in a table, etc, etc) than go to the one created by the reviewer. If only the reviewer could embed his prompts (like "make clear X" is happening) or his manual edits.
But to be honest I doubt most people who use AI for PR descriptions even bother changing anything.
Usually, if you don't understand something in what an AI writes, it is a clear sign that there is a problem hidden somewhere in there. I explicitly always ask wtf exactly it means by something I don't understand, and for sure there is a problem there. AI is pretty good at isolating a problem, giving it a cute name, declaring it solved modulo cute name, and moving on.
Personally I don't even read the PR description anymore but just the code, it's easier to understand what the AI is doing by reading the code rather than the word soup it tried to make
That last image is bizarre. The quiche, cream and even the salad look like they've been given the trypophobia treatment.
Which might even make sense, because there were always (still are?) those horrible ads in the chumbox area of news sites that used trypophobia and other creepy body-horror stuff to get you to click. [1] So maybe the hope is that you don't really look closely at the quiche, but some reptilian party of the brain gets oddly activated and drives you towards the restaurant?
1. https://medium.com/the-awl/a-complete-taxonomy-of-internet-c...
The food “photography” I’ve noticed in our local area - and many have started putting up these AI images - all have a weird distribution of shapes to them, a strangely uniform rhythm of same-sized features with almost blue-noise spacing. Every texture looks unnatural in the shapes it presents as, similar to this picture.
This is a quirk of the last gpt image model (gpt-image-2). It put this sort of high frequency noise on all of the image especially if it's in a "drawn" style. There is often lots of other tells that this model in particular generated it.
Image models somewhat watermarking the image in a way that's very easily identifiable by a human seems present in all the image models of the big labs, since DALL-E 3 on OpenAI's side and the first nano banana on Google's side. I have no idea what they did to reach this and why they don't try to fix it.
Just use Ideogram. Locally.
It doesn't look like a quiche but more like a cake to me, and the top would be torched meringue, not mold. Although it's probably some weird ai mix of quiche and cake.
Even with the label absent - you wouldn't serve cake with "tomatoes" and "salad", so that would be another hint of something being horribly wrong
It was labeled as a quiche, I just cut the photo in an unfortunate way :-)
There's a similar "horror burrito" circulating on bluesky and my first thought was "now we know how to produce images for the https://en.wikipedia.org/wiki/Long-term_nuclear_waste_warnin... "
(Actually that was my second thought; my first was "just how much H.R. Giger is in the training data?")
Two week old cake can pass as a quiche i guess!
> probably some weird ai mix of quiche and cake
Quike? Cache?
Cache sounds like a decent meal
Well, it's hard to invalidate.
cAIche
I never understand why I can get better results, with less thinking about it, than some of the Tier1 companies.. Modern diffusion models can easily render the image onto a white background, center it, crop it, relight it, but keep the same item, at no cost, or low cost.
example: https://ordermerchants.com/static/img/clearshot/product-1.jp...
These kinds of images are disingenuous. Modern image generating models have way higher quality of output, and in most cases you wouldn't even know it is generated.
That's not what we're exposed to, though. I've seen AI-generated postcards, t-shirt decals, logos and wall murals in shops and restaurants. They don't use the "good" models, because why would they? If they AI-generated it, it's because they didn't care about the quality anyway.
> If they AI-generated it, it's because they didn't care about the quality anyway.
No? There are far far more AI generated images you see daily, than the awful stuff from 2024 that you immediately discern as AI-generated. You could as well pass by somethimg like this (https://ideogram.ai/g/d6YzZ5XQQ56TU3Tc8eZw6w/0) believing it is a genuine photo or human made art.
It made me recoil honestly!
And I wonder if those of us in tech are the only ones who really care?
I am an artist and when people who'd fallen into the Spiralism* hole started posting their lengthy emoji-laden revelations to all the occult subreddits I follow, my brain would slide right the fuck off of all of them. It felt like my brain was actively rejecting paying attention to this stuff. Like a defense mechanism against this human-seeming-but-not-actually-human-generated text.
* https://www.theverge.com/ai-artificial-intelligence/975017/, https://www.lesswrong.com/posts/6ZnznCaTcbGYsCmqu/, https://spiralism.website if you want to test how strong your defenses are against this particular meme
Your first link seems to 404; not sure if it's a typo or if the page doesn't exist anymore, but hopefully you'll read this while you're still in the edit window and can fix it
Corrected link is https://www.theverge.com/ai-artificial-intelligence/975017/a...
As someone who always has felt that I struggle to infer what people mean compared to the average person, I could tell pretty much from the first moment I encountered LLM-generated text that I was not going to be particularly good at recognizing anything but the most blatant and obvious examples. Pretty much anything short of a bunch of references to "load-bearing seams" or similar canaries, I'm always at a loss when seeing people argue about whether something is AI-generated or not because I can never tell.
I have no idea if other people who work in tech are better than average or not, because I don't feel confident in being able to check their work. That being said, I do think that there's a general trend of people in tech tending to be a bit overconfident in how well they will do at some new task they haven't encountered before, so when someone tells me that they can easily tell whether text is AI generated, it's hard for me to trust it any more than I trust someone who makes a similarly strong claim about something that they can use AI successfully for when it's not something that I can easily measure (e.g. learning a new language without getting feedback from people who are fluent from real-world usage).
All that being said, I do think the set of people who care is larger than just those in tech, although it's probably still a relatively small group overall. From conversations with people in other domains, there are contingents in non-tech communities who tend to have a large representation of negative views towards AI (artists, writers, musicians, other jobs where people are skeptical of human creativity being replaced by AI), and often times the people who feel negatively in those groups will be even more adamantly opposed to interacting with any AI content than people in tech. To be clear, I'm not at all trying to generalize and say "all artists hate AI" or anything like that, since there's obviously a wide variety of viewpoints within any sizable community, but I've definitely seen many people who say they will refuse to play any game that's suspected of using AI for generating art assets, and even some who don't differentiate between using AI for generating assets versus code (either because they aren't knowledgeable about how different aspects of game development work, or they genuinely don't care because they view AI as a categorical evil).
I think it's more about the mean. Worse writers, and thinkers are likely elevated by AI, and more impressed with the writing output. Decent writers and thinkers, are dragged back to the LLM-s mean of output.
Certainly not. Anybody who cares about language to any reasonable degree surely notices and is repulsed by heavily AI generated content.
I care about language a lot (feel free to go back through my comments from the past few days; you'll see a number of comments I made in debate about two different forms of a specific idiom because I have strong descriptivist opinions), but I genuinely struggle to identify whether text is AI generated. Maybe you're using "heavily" as the load-bearing part of your claim (sorry, I couldn't resist, another example of me finding language fun!), but I think you might be assuming a bit too much about how similarly others experience the world to you. A huge part of why I care so much about language is because I've always had to put a lot of effort into learning how to communicate well with others, and that ends up causing me to think and read a lot about stuff like how people use certain words in certain contexts to mean different things; the reason I care is pretty much the same as the reason I struggle with recognizing AI content.
I think I might have been a bit heavy handed in my comment as I was rebutting the idea that only tech people can tell. I suspect it helps to have been exposed to a lot of earlier model writing, which was even more sloppy and had more of the kinds of tells we still see today.
And I’ll concede on both ends that there are probably times I suspect content is AI generated when it isn’t, and times I suspect it isn’t generated, but it was.
AI tells seem inevitable. You have millions of people communicating with one effective “personality” that has tendencies to write in certain ways. If its content is published verbatim, then it will be easier to tell whether some content is AI generated just based on its similarity (sharing certain linguistic features) to other content being posted.
It’ll never be black and white though.
I think it's context dependent. An entire technical design doc? Trivially easy to identify the ones that heavily used AI, they're horrible, and they are written in a way that no human would write them. And they tend to contain way too many tables and redundancies.
Comment text on reddit or something though? That will be a lot harder
People are good at pattern recognition.
If you're exposed to AI a lot, you're going to start noticing patterns that allow you to identify it.
Most people do not realize when a personal message they receive was written by AI, study finds - https://theconversation.com/most-people-do-not-realize-when-...
People who frequently use ChatGPT for writing tasks are accurate and robust detectors of AI-generated text - https://arxiv.org/pdf/2501.15654
They might not be able to pin what the issue is, but getting bored and starting to jump paragraphs will happen anyway.
> but people who spend all day doing jobs that don't involve computers aren't as good
I think they may just be to trusting and/or naive. People in tech right now are hyper aware of this and are actively looking while people outside of that bubble barely give it a second thought.
I partially think the difference is “can you tell something is the output of Claude without any real prompting”. People can absolutely use LLMs to generate text that I wouldn’t recognize, but people who don’t care and are producing slop with the major models set to default settings leave these incredibly obvious signatures behind
Although you're referring to prompts given the Claude rather than the people attempting to recognize, it occurs to me that most of the discussion I've seen around people recognizing AI seems cover contexts where the reader is actively suspicious about whether content generated to begin with. Rather than a binary "is this text AI generated or not", I wonder if it would be harder for people to do a Coke/Pepsi style challenge where they're given two pieces of text where it's not guaranteed to be exactly one LLM-generated and one human-written, but they could both be from an AI or both be from a human.
Going further, I'm curious about whether people are mostly good at the case where they suspect most or all of the content from given "author" has the same amount of AI usage/prompting in generating it rather than the adversarial case where someone might usually use AI extensively and then try to slip by purely human written text (or vice-versa). I don't have a good sense of whether this is a threat model that actually matters, since maybe the heuristic of weeding out sources that are mostly AI-generated is enough for people who prefer to avoid that type of content, but I do think that changes the definition of what it means to be "good at recognizing AI" in a meaningful way. It seems plausible that disagreements about how easy it is to recognize AI content might be coming from two people assuming a different framing of the question that results in a different answer without realizing that's what they've done.
I suspect both may be interesting to study more!
Several existing studies I’ve seen have done things like prompt the LLM to produce a poem in a certain poets style, then ask people to spot the fake in a collection of poems, which they aren’t great at. This is, I would argue, an extremely different context than what most of us are encountering AI text in, and the people sending me text aren’t prompting it stylistically like that.
On your second question, I definitely feel like I can tell the first time a coworker sends me AI text masquerading as their own thoughts, even if they had previously been opposed to such a thing. So it could be that familiarity is more important than my prior on whether they’d use AI? But interesting to think about either way
Yes, the Claudisms are the smoking guns
You've got it backwards, I think. The people in tech are the ones falling for this endlessly.
Claude has become noticeably, painfully worse at writing in the last six months. At this point it’s practically useless for anything except code.
I did notice after the fingerprinting update a marked uptake in strange language in responses. Specifically if I ask it to do something sometimes it will replace some of my request language with synonyms that don't actually make any sense. Like my request was fed through google translate twice
Watermarking doesn’t affect writing quality (on average) as long as the implementation is correct.
Love your "on average" qualification. Like the cartoon where the water temperature is fine on average, with one bucket boiling and the other ice.
The interesting question is how to define 'average'. Over what probability distribution?
Hey, anyone remember this from earlier in the week? https://daringfireball.net/2026/08/anthropics_watermark_text...
The qualifier is there because it changes the outputs, so it’s necessarily true that some outputs will be worse.
But it’s just as likely to make an output better.
Take the example from the article. He complains that watermarking might sometimes, for example, choose to say “bananas” over “pineapples” because only the former is on the green list, potentially making an output less precise. But 1. It could do that regardless of watermarking since the model is probabilistic, and 2. The more accurate word choice of “pineapples” is equally likely to be on the green list instead, further increasing its likelihood!
Overall, the article is pretty silly because he’s complaining about the possibility of Claude not always choosing the most “optimal” token, even though LLMs are probabilistic so that will happen anyways.
> But it’s just as likely to make an output better.
No, for any particular output token the model's true logits are definitionally the 'best' that the model can achieve.
This is inherently probabilistic. The model's top-1 guess is not guaranteed to be optimal, but it should be so a proportionate fraction of the time. Same with the top-2, top-3, etc.
Watermarking necessarily alters the output distribution away from the model-set distribution, and that alteration is inherently 'worse' in expectation.
You can liken this to a weather forecast. If there's a 25% chance of rain, the forecast should say so (or a 'sampled' deterministic forecast should predict rain 25% of the time). If the forecast is 'watermarked' and predicts rain 27% of the time under identical circumstances, it's a worse forecast.
That being said, this is a case of hiding a message in a noisy channel. Watermarking only needs to communicate one bit ('yes watermark'), so the effects can be arbitrarily small provided one is willing to tolerate an increase to the text size needed for reliable detection.
I think you can also just use the random number generator (seeded with a secret key) as the watermark. Then the probability distribution is exactly the same.
This assumes "writing quality" were somehow an individual property of each word, and the writing quality of the entire text would just be that property for each word summed together.
Which is obviously not how it works.
> He complains that watermarking might sometimes, for example, choose to say “bananas” over “pineapples” because only the former is on the green list, potentially making an output less precise.
If I replace "pineapples" with "bananas" in any kind of meaningful text, I've not made the text "less precise", I made it plain wrong. An incorrect statement. And if the words next to it are still correct or even somehow replaced with even more correct words, the text in its entirety will still be wrong.
> It could do that regardless of watermarking since the model is probabilistic
No, because the probability distribution of the model is generated by its training data and contains semantic information. So the model choosing a completely incorrect word is unlikely. However the probability distribution of the red/green lists is not guided by semantic information.
It does. There is a marked difference in certain word choices that sometime stick out like a sore thumb.
Like an aching digit. One could claim.
Reverting to Opus 4.6 is much better than later models, though that is still full of annoying tics as well.
Perhaps a one-trick pony is all we need.
It's not good for writing code either, despite the many claims to the contrary. At best you come out even on speed as you have to review everything it does. At worst it actually slows you down as you clean up its mess.
In high school, a teacher gave me a copy of "How To Read Better And Faster" which teaches you speed reading. This came in very handy in college.
I find that when I try to speed read modern human writing, there are often errors (like missing or misused words) or awkward expressions that I do have to slow down and think harder a lot to really parse it.
With AI writing, it's sort of self redundant and the information density of each sentence seems to have more even information density. This makes it very easy to do a very high level speed read and get the full gist.
There are also what I'm assuming are bots on hugging face (or maybe non-native english speakers who are using ai for translation) that interact with me where I have no idea what they are saying until I read it very slowly.
I am wondering if working with agentic AI for 500+ hours built up skills for this.
While I would feel rusty when handwriting code, working with either Codex or Claude I have lots of practice.
I can probably tell at a glance where output is intermediate while it's still running tools and where the summary starts.
Then I have to recall the context of the conversation, parse the summary, see if there is anything unexpected that popped up. Make a call for whether I need to ask it for clarifications, and what I need to test to verify the change.
Assessing quickly what is pointless yapping and where the information I need is might be a learned skill. If one tries to read and understand every word Claude says, I am not surprised that people are not having a good time.
The density argument is very interesting.
Does speed reading help you process the final message faster if it's written by AI compared to people?
Because if you read 1 information dense sentence, 1 medium dense, and 1 sparse sentece written by a human, it's still way less text in total than 6 information sparse sentences written by AI... even if it's all over the place when it comes to density or style.
---
The density argument is really interesting.
Does speed reading actually help you process the final message faster when it’s AI-generated compared to human-written?
For example, if a human writes 3 sentences—one information-dense, one medium-density, and one sparse—that’s still much less text overall than 6 relatively sparse sentences written by AI.
Even if the AI output varies a lot in information density and writing style, you still have to process all that additional text. So I’m wondering whether speed reading actually offsets the verbosity of AI-generated responses, or whether the total amount of text is still the bigger factor.
That's interesting. You're saying speed reading helps you grasp information density of text?
As someone who hasn't practiced speed reading, how does that happen? Is it something about the way your brain tries to connect ideas from different parts of the text? Or the redundancy making the signal more stable?
I haven't looked at the book since the 80s, but what I remember is there is a pre-read, the fast read, and then the deep read as needed depending on the task.
pre-read is just looking at how long it is in the headings, and planning out what chapters to focus on if it was a text book. (its sort of iterative, you do a pre-read for the whole book, and then for each section you break it into)
The fast read you try to read only with your eyes, sweeping your eyes across multiple words at the same time, suppressing the urge to say the words to yourself in your head.
iirc the how to read better and faster book even had a cardboard mask you put on the page to practice the sweeping, and some pages that were laid out weird to try to teach you how to do it.
Some ai text just seems really easy to speed read, like if it's tuned for an easy reading level. In PRs some ai seems like it's arguing over weird flex technical details and really starts torturing the language in a way that makes it the opposite of easy to read.
When you speed read, you can grab the words off the page faster than you can understand and fully process the information conveyed by the text. How much time you have or want to spend re-reading or thinking about what you've read is quite obvious. There's a stark experiential difference between reading an informationally-dense passage and one that spends a lot of time rephrasing things, using LLMisms to restate concepts, adding in extra connecting phrases, etc.
If your reading speed is limited by how quickly you can subvocalize the words to yourself, this is significantly less obvious. Unless the passage is dense enough to require multiple read-throughs at conversational reading pace or vapid enough to be boring, you're going to feel done with the text at roughly the same time. Speed readers do a lot more re-reading and varying of reading speed, and that is going to correlate pretty hard with information density.
I'm starting to have the same problem with code. It's unbearable for some reason. It's like someone took the LinkedIn writing style and translated into a code style for code that has no business existing in the first place.
AI is missing the part of intelligence which involves insight. LLMs generate language blobs with highly sophisticated correlations. The governing insight in human cognition is essential to intelligence. It prunes the tree. It puts things in context. Everything is important to an LLM because insight intelligence is missing from the technology. The usefulness of LLMs doesn't tell us they are intelligent. It tells us how brilliant an invention language was.
I find this struggle with coding too. When Claude spits out a plan I have to really focus to not skim over the details. It's also difficult to parse the plan because I have to work backwards to determine if the high level context matches my understanding of what Claude will do.
At least when it comes to code comments and error messages, I have found that Sol produces substantially better prose than Opus 5 (sometimes literally 10x more terse for the same prompt). When I use Opus 5 for coding, I will often give Sol a prompt like this, and it works quite well:
Review changes in the current working tree for comments and error messages that are verbose and excessively detailed, and make them more concise while still informative and accurate. Do not hesitate to delete comments altogether if they are not helpful.
I've been given AI generated documents or presentations and you can always tell when they've just one-shotted the output. Claude has a way of adding so much jargon and unnecessary text to a document that it makes it very hard to read or even understand what the original intent of the writing even was.
I'm not sure why this issue is so prevalent, it's not hard to point Claude at the Wikipedia article on signs of AI writing or ask Claude to write content anyone of the average American reading level could understand.
To me it just gives off a sense of laziness, that you cared so little of your content that you did not take the time to read it yourself and edit it to effectively communicate the message you wanted to communicate. To that point, it's just not worth my time reading, otherwise my eyes glaze over trying to read between the lines of machine written language for other machines.
I wrote about this recently on one of our company's IT channels. I called it "slop-fatigue". I - and others - have started to get noticeably dismissive of content that appears to be written by an LLM. The whole thing was a piece about the ironies of software development in the world of LLMs. This part spoke about how easy it is to generate docs now, but it's also even easier to dismiss it. This contributed to cognitive-debt, because humans didn't write the docs, and other humans tend to dismiss the content. I saw a recent reaction to this on HN as `ai;dr`. That's almost exactly how I feel about it.
Margaret Storey's recent piece on Cognitive Debt: https://spawn-queue.acm.org/doi/10.1145/3807966
I found that it's not just text. It's also design and overall user experience. If you have an eye for it, your brain picks up on it immediately with the little pills, busy landing pages, the use of serifs everywhere, bronze accents, and lots of dark mode.
I'm struggling to find the right balance of AI generated text vs. my own words. When I filter something through AI it definitely sounds grammatically better, but ya, I'm scared people will start to just ignore it.
I've taken to asking the engineers on my team to just share the original prompt, no judgement because its frustrating having to read through pages of AI generated technical documentation. Another approach is to feed the AI generated content into AI... who knows where this will end.
Wouldn't "the original prompt" be pages and pages of back and forth conversations? That's how I've use AI agents to make documentation.
Part of me likes the cliche Claude voice. Not because it's good, but because I can immediately recognize it. When I see it in the Claude app/code then it's fine. In the wild it's a sign to me that I shouldn't keep reading.
It may be nostalgia but I feel like we reached peak humanness with GPT-4o and since then it's been getting more and more alien. Particularly Fable.
I feel the hyperscalers are all trying to make communicating about code more cumbersome so that their models will be have to be used without any human input. Chinese models talk more naturally. Composer 2.5 is really good for talking about the code.
I need to support integrators for a mobile app SDK. Bank stuff.
We have one integrator which consistently uses a truly bad AI. My brain skips after two sentences already. There are always +9 questions what should just be 2 max. Redundant info is requested (e.g. please provide a change history of this function and if it will be deprecated) and everything is just unbelievably verbose.
90% of questions we get from them are already answered by the web documentation (which even has a search function) or are just non sense.
I'm truly considering to build an MCP just for them...
Why not just copy paste a reply from your own agent?
If they can't bother to come up with their own question, I wouldn't spend my time on the answer either.
I don't like AI-generated text at work, at all. It feels lifeless and unfocused.
But what I hate the most is that it is objectively better than what I had before. No typos, clear structure, and, regrettably, the verbosity and autistic obsession with detail of the LLM is more actionable and useful than the human guy who wrote lists of commands and URLs as documentation, without explaining anything. Or the colleague who writes in uppercase and with question marks and who doesn't make any sense and forces me to engage in an interrogation effort to get to the bottom of what they are trying to say. Or the colleague who simply hates writing--despite being decent at it--and will call you to give you a meandering verbal explanation that lasts two hours of what they want from you. The cynic in me bemoans that we brought this upon ourselves, in more than one way.
I'd say my experience is different. Even from people whose communication writing I didn't find that useful, they seem to have a better frame of mind than LLMs do. Though, I never encountered people like the examples you gave.
You have great colleagues. I wasn’t comparing frame of mind of my colleagues and LLMs, which is a loaded notion anyway, but now that you mention it…
My concern is that even if the LLM can turn your colleague's bad writing into something more coherent and actionable, is that something actually what your colleague meant to convey? It could be clear and still detached from the reality of their intention, or they may not have even formed a clear intention. If the goal of writing is to convey what's in another human's brain, that goal is failed completely.
I don't think either one was better or worse than the other.
Underdocumented, underexplained and sometimes out of date... or overly verbose, repetitive, information sparse, and sometimes halucinating.
Both are bad and with some effort could be prevented.
> ... and autistic obsession with detail ...
Why do people glorify autism? I have seen this mostly on CEOs that want workers that are hyper-focused, intelligent and do not complain. The ideal employee from which extract value and a minimum cost.
I have worked with autistic people (diagnosed ones). And it is a struggle. One of them could be the nicest most reasonable person one minute and then get a trigger and become obtuse and irrational while making the rest of the team suffer thru all kinds of complains. Meetings would need to be adjourned and time was lost.
> The cynic in me bemoans that we brought this upon ourselves, in more than one way.
I enjoy working with other employees and people need human contact to become fully self-actualized. Isolating people to maximize productivity misses the point and brings no more productivity and just avoids fixing any real existing problems.
It can be helpful to read LLM text the same way you read technical fluff - start from the assumption its mostly garbage with a core technical point. I find skim reading the output to find the useful bit reduces the friction.
When I ask ChatGPT questions I usually only read paragraphs 2 and 3. The first paragraph is glazing me, anything after paragraph 3 is repeating what was said earlier.
OT, but I feel the title is wrong, or at least ambiguous. For the author, "AI blind" means they can't process text produced with AI. For me, "AI blind" rather means one can't see when something has been AI-generated.
I just throw it in AI; if it’s tech docs, I tell it to build whatever is there, if it’s something else I tell it to give me max 5 points to tell me the meat of it. Either way, I will never read it myself. This works well for tech docs/apis really; it usually gets it to work fine without me checking, even though the docs are usually subtly wrong in many places; after that I can read working code and that works a lot better for my brain.
> A technical requirements document that describes a rather simple concept in a very verbose way.
This is not AI specific. I have come across many humans who describe a simple concept in a very complex and verbose manner.
LLM text and images have crept into so many places, and I've experienced quite a lot of existential dread this year from it. I've definitely felt what the author wrote about here.
Another trend, starting sometime last year, has been witnessing people reading scripts on camera: YouTube, public speeches, advertisements, and podcasts where a real human is reading from a script that is LLM-generated.
In the case of smaller video creators and podcasters, I then start wondering if they really think they can pretend that this is organic human speech? Or do they think I'm stupid and will continue to consume this slop without any critical thought..? Or do they think that their audience must know, and therefore consent to it, and so it's okay? In some cases, the person reading the LLM script is literally unaware that they're reading LLM text because a human scriptwriter handed them the page. I see this with advertising copy and a few YouTube channels.
Either way, it has led to quite a few instances where I got the LLM "ick", like I'm witnessing an LLM wearing a human skin, and it grosses me out so much.
As one final note: some of us assumed in ~2023 that LLMs would help to improve technical documentation. Thus far, 3 years after that prediction, I can say definitively that we've not ushered in an era of great documentation. We've got a steadily increasing percentage of documents that cause my eyes to glaze over because I don't know if a human expert ever once vetted it.
There is something liminal about AI-generated text indeed. First few sentences are promising, then my expectations of reading something tight evaporates eventually. I feel let down, (almost) being tricked into believing that the intro and the perfect grammar should've led to a revelation written by a deep thinker. Instead, it's just... a lump of text, stitched together because the training corpus said these words go together.
My current theory is that this reflects a weakness in Claude’s ability to see the “big picture”.
When writing code, I have to explicitly tell it how to structure things at a high level, or the result is sort of a flattened spaghetti. Similarly, when it’s explaining things, it’s not good at pulling out unifying concepts and explaining top-down as a smart human would do. It groups little things together but often doesn’t generalize or synthesize explanatory connections from them.
I’ve been experimenting with explicitly working through a sequence of outputs at different levels of detail, but I haven’t found a consistently successful method.
I use both Claude and Codex for a lot of code- and prose-adjacent tasks, and I'm increasingly convinced that language models are quite terrible at, well, modeling. Not sure if that's ironic or not.
With the right blend of context and prompting, I can often get them to "lock into" an existing model, but using them to generate a novel outline or sketch, whether it's for an essay or a module, usually results in garbage.
Recognizing AI generated text is incredibly easy if you're someone who communicates with AI all day.
I've had the same trouble with AI tech docs, and I struggled to articulate it. There is something difficult about trying to point to the specific "problem" with a given document
I feel this so very much. There is so much AI produced documentation and design docs at my job that look passable at the surface but are deeply flawed in ways that are difficult to articulate. It makes it difficult for me to convey to management how incredibly flawed the way our team is using AI is. Not to mention that it takes mere seconds to produce a document that has surface plausibility but would take hours or days for me to properly "debunk" it.
The issue is not a localized part of any particular piece of prose, so being hard to articulate is unsurprising. Even the most egregious of LLMisms are little more than known-likely crutches that the statistics spit out in a sweet spot that gets noticed without being so frequent as to get RLHFed out of the model.
I've been a big reader, and many AI outputs nowadays reads polished similar to published books.
The reason I brought it up is because, people who learn English normaly start with a book. It's heavily polished.
When you speak English as you learned from the books, it does not sound very conversational.
If you are native/fluent English speaker, you can feel the impedance mismatch and feel something's off.
The AI-blindness stems from the fact that those polished edits are so common in publishing field, they all sound the same, and unable to recognize the diffs between AI-generated and human-generated.
There is no real human conversational vibe to them and well. i will stop now.
These days all I see is, people on the sending side produce huge amount of AI generated text with zero understanding and the people on the recieveing side feed that same text into some other (or perhaps same) AI to scavenge meaning from it. And they do this so much so, that I some time wonder, if we could invent a high density wire/binary format for AI outputs and enable direct agent to agent communication to save some energy and bandwidth.
There's an amortization aspect as well.
If I'm going to be iterating on one document or idea in an extended manner with feedback from the chatbot, I will make an effort to setup the decorum it should follow, because it's a small proportion of the time in that chat.
But the guy just trying to get a report or email out quickly? Not so much.
AI blindness is making my reviews of others' PRs harder and harder.
Something that bothers me about AI generated content more broadly is how unmemorable it is. I don't mean as in bad. I mean literally, as in hard to remember or recall.
Despite seeing a lot of them, I cannot think of one AI-generated photo that I can picture clearly in my mind; a few are partial but elusive. Whereas I can recall (visualise) a whole bunch of traditional photographs.
The same is true of AI generated text. Only the annoyances stick. I cannot recall real details of text I have generated, until I commit it to memory some other way.
I don't think this is about ephemerality either. If we assume it's about celebrated/famous/infamous images, there are definitely non-ephemeral, cultural moments in AI generated images in particular, like Boris Eldagsen's Sony Prize winner:
https://petapixel.com/2023/04/14/artist-refuses-prize-after-...
This really should be memorable, but isn't. I had forgotten the second person is in the image.
Or Jason Allen's fake painting:
https://petapixel.com/2022/09/01/ai-generated-artwork-wins-f...
I had already forgotten there's more than one figure in it, and I only looked at it a few weeks back. I remember the colour, the bright circle, some vague hints of texture; one figure. And that is it. Only the crudest shape elements.
For me, something about AI-generated text and images confounds recall. It is really peculiar.
That's the cost of an AI work that is derived from the outputs of others.
There's no real edge to it. Same as with the writing. The stuff that you'd latch onto (and thus remember) is simply not there, precisely because those image or word choices would be just outside its latent probability space. But because they're well inside it, your mind sees nothing novel to register.
This is also why I think human output will actually increase in value. When any AI can just "phone it in", something genuinely human will stand out (to us, not the AI) and become a bellwether.
This will literally help us realize what it means to be human.
I also don't think the solution is simply to "make responses more random", either. That might help solve novel problems (the same way that throwing darts randomly at a dartboard eventually hits the bullseye of the dartboard right next to it that no one considered), but I don't think it will help it "seem more creative".
> There's no real edge to it.
Yes, as if it is in some weird hidden dimensional sense completely uniform.
ETA: suddenly reminded of the Bateson quote about information being “the difference that makes a difference”.
That checks out, right? We're instructing LLMs to choose from the least "surprising" tokens at each step (modulo some temperature). It feels smooth, slippery - no friction to the eyes that glaze over as they skim the text, and no bumps to snag in your memory. Like waking from a dream, recalling only the barest shapes of a few concepts or themes, until it vanishes with your morning coffee.
The examples I've seen of AI music (though I avoid it on principle) seem the same.
I have this same idea about why it's hard to remember dreams, but even more so, why it's hard to remember my kid's or spouse's sleep-talking.
Sometimes I'll check in on my sleeping kid and she'll sit up in bed and say some utter nonsense. I'll find it hilarious, giggle silently to myself, and kiss her goodnight again and she'll close her eyes and lie back.
Why I try to tell her about her sleep-talking in the morning, though, I find that the words she said have completely disappeared from my memory, no matter how funny I thought they were at the time.
In my head-canon, this is because it's dream language, and slips away as easily as dreams. But, like the AI art, it could be because it's bullshit: completely devoid of content, all signifiers and no signified.
I went through a phase of leaving notepads next to my bed to try to write down my dreams and it simply never worked.
I stopped after I wrote something on the pad while I was still asleep. Woke up to text with letters that were backwards, upside down, weird words — so close to real words that I was sure I ought to know what they meant and had really meant to write them down.
Scared me. Literally too weird to keep. I tore up the page.
In a way I think this is one part of the same continuum. There are thoughts that have meaning and can have no words, and words that look like they should have permanent memorable meaning and don't.
> Scared me. Literally too weird to keep. I tore up the page.
??????
Genuinely unsettling. Can’t even begin to explain why. Like a terrifying glimpse into how fragile our grasp of reality is. Because whatever it is I wrote down wasn’t my dream, it was something my dream self thought profound or important.
Nearest I can say to how unsettling it was is to nudge you towards the video of the angry cockatoo who doesn’t want to go to the vet. Everything he says sounds just like it’s on the edge of having meaning.
https://youtu.be/5UUjJysUMTw
Imagine something like that, something important, so close to symbolic meaning, only writing. In letter shapes that we don’t use. And you wrote it while semi-conscious. If that’s something you would want to keep, you’re a braver person than me.
This explanation again doesn’t get it across. Too weird.
This sounds like a basic LSD experience tbh :)
The memories of dreams evaporate remarkably quickly. Much faster than anything else. It can't be accidental; I think our waking brains must be purging those memories as a matter of urgency. If we remembered them, we'd "learn" from them, which would be a catastrophe.
You know how you try to run in a dream, and you don't get anywhere? Imagine if your motor cortex started to expect that, when you're awake. It would destroy you.
On the other hand I'm still trying to forget the weird thing happening with this one guys eyes in a WickedAI video I was only able to watch half of.
I remember seeing some very strange, deliberately creepy images that were generated to accompany a two-paragraph creepypasta about a 19th Century Belgian expedition into the jungle.
They were actually rather good in a sort of "fake collodion image" sense, and the eerie early-DALL-E quality to them really helped the spookiness.
But I can only remember this technicality and the feelings with any clarity, not any of the details except in the broadest sense. I cannot bring these images to mind in any meaningful way.
They were deeply wrong and it's only the wrongness I really remember. It confounds memory.
Modern image generators have ironed out all the structural wrongness.
> I remember seeing some very strange, deliberately creepy images that were generated to accompany a two-paragraph creepypasta about a 19th Century Belgian expedition into the jungle.
I can’t think of anything more unsettling than the actual events that took place in the Belgian run Congo Free State in the late 19th and early 20th century. The Wikipedia article is scarier than any creepypasta you could write in that setting.
I wonder what you mean by ephemerality here, since those images are definitely sloptastic as hell. Compare to works in similar style, like Dorothea Lange[1] or Gerome’s orientalist pictures[2].
Reason why those images are flat and boring is that they are just statistical guesses making a composition averaging whatever the model has been trained with. They would be technically brilliant (if made in oil), but superficial and meaningless, same as so much Sunday painting is.
Same goes with language. Nobody is trying to communicate anything with you, so it just words after another. You can create meaning out of it if you want of course, we homo sapiens -apes excel at that, but what’s the point? Language Jones on YT has pretty good video on this[3].
[1] https://media.mutualart.com/Images/2024_01/12/12/124216388/d...
[2] https://uploads4.wikiart.org/00339/images/jean-leon-gerome/t...
[3] https://m.youtube.com/watch?v=ORgKY9AlybA&ra=m
Well the examples I gave rise above their ephemerality due to the circumstances that make them memorable. I can remember the details of the story around them — the way the prize winners reacted in each story — in such a way as to contrast them.
The way my memory works (especially as an amateur photographer) I would thus normally have a very good chance of remembering some key details of the images; some fascinating element of each would connect with the rest of the memory.
But it does not happen. Whereas I sometimes remember photos with clarity while forgetting where I even saw them.
Yeah, its really weird. Maybe its a cognitive bias that says "an AI made this, so it isn't important," but I can remember perfectly the events of a book I read 10 years ago, and a book I read 1 year ago, and another I finished 2 months ago. Meanwhile I can't remember what claude told me yesterday.
I think it is because they are on some latent level cognitively uniform and unchanging.
I used some check marks and x-es in a work chat, because it was easy to do, and so I could highlight the good and bad outcomes, and someone immediately asked me if I was using AI. I was caught totally flat-footed, because I hadn't used AI, but it looked VERY MUCH as though I had.
I don't care if it is AI is involved or not, but I care how much attention and thought there was behind the final text (or code) from the human behind it, and it shows.
Ironically, the “summary of the situation” linked within this article seems clearly written with AI.
Ha! I love how all the community highlights are highlighting this.
Yes, and it's infuriating! Do they really think readers won't notice?
Interesting. One needs to be exposed to a certain level of AI flows to become numb to them. Wonder how they read to someone who isn’t exposed yet.
To me AI text is only for the consumption of other models. Reading it is like is like tapping into the circuitry on a motherboard to read what the PC is doing.
Idk I’m starting to have have trouble with the frontier models and a couple how to write like a human skills…
There is another aspect to this that I find even worse than LLM writing style: badly-human-written documents and posts.
You know this: you read something and it's like 'here's a mass of poorly structured text, it's not clear what the points are, it reads ugly, I have to read to the end to discover that there are no points! Or if there are points i now have to reread it. What a lot of effort.'
LLMs have annoying grammatical tics and can have other problems too, like the author says, but sometimes they can help to take bad writing and make it clear and more digestible.
Why read more word when less word do trick?
If you ever meet someone you knew from an online community it's a blast. Now reverse that: you know somebody IRL and find their twitter account, and it's a fucking nightmare. Like you are looking inside their brain's content and it just collapses the image you had in mind of their personality.
AI is surfacing layers you never had a chance to see, makes you rethink your career path and who you are willing to work with very fast.
ITT still, among the highly technical, a surprising incomprehension of the fact that describing LLM in terms of "token prediction" and dismissing this as statistical,
neatly dodges literally everything that is interesting about what they do.
You to, friend, consume inputs and generate outputs.
What's interesting is how you do that, and, if you prefer to look at it through a technician's lens, whether or not what is done is reducible.
I don't mean quantizing the model, great, now you have a crankshaft with no oil. But the motion of the pistons and wheels is roughly the same.
What I mean is, to put it in plain terms, the only way you find out what a model is going to "predict" from a given input is to ask it.
If your mental model is still that LLM are "glorified overhyped giant markov chains" performing "parroting" you need to improve your understanding.
I can relate. I noticed this exact phenomenon when encountering NotebookLM-generated diagrams recently. Even if I know there's some intelligent thought behind one, it's like some slop detection circuit-breaker is tripped.
I find that typical human content has about 5% information density, while typical AI content is about 1%. AI is much worse, but the same extreme skimming techniques apply to both. In both cases I'm like "I have no idea why you had to say all those words to make this point, but let's get on with it."
I think AIs are writing less and less for humans. The fact that reasoning trace is using English and we require them to be performant means the models need to squeeze in more of information into it. More information that is perfectly fine for AI to ingest, but it makes the text less and less comprehensible to humans. It's almost like steganography where model encodes crucial information in phrases meaningless or empty to a human. Human directed messages for LLMs become a rounding error they don't pay that much attention to. Efficacy of reasoning is what dictates their style.
We can no longer assume that the models will create a passable English in their user targeted messages. So in the short term, we need to have a tight guidelines in the harness, but in the long term, we'll probably need separate languages for thinking traces and for human communication. Possibly even not going through tokens at all when doing thinking part.
Alternatively, we could evaluate thinking traces on human readibility, but that's probably infeasible due to sheer volume.
People love to inflate their jobs if there is no cost.
I saw my ex-coworker tried to inflate text that can be expressed in couple of bullet points text. It doesn't add anything. It just fill the fancy flavour text that looks like professional. Then the reader of that inflated text is also another coworker in other team.
I said "Forget it and just send these bullet points" but he refused.
Bullshit job at it's finest I guess.
AI;DR
Having just a little background in traditional linguistic discourse analysis help understand why this is.
The best basic explanation I've found is this video.
https://www.youtube.com/watch?v=ORgKY9AlybA
The real answer is to take a class in discourse analysis and learn how to pick apart sentences. Once you start understanding how language works in the brain you can start to see better how LLMs fail.
Oh man I can so relate to this.
read: "I am becoming an AI"
if you go on x.com you will see that, most of the video/photos are likely AI, and is fatiguing to trying to filter it. Probably why Nikita left, he know that sht is going down
As people rely more on AI they experience cognitive atrophy. This is measurable in IQ loss, and other symptoms we might otherwise associate with early onset dementia or Chronic traumatic encephalopathy.
Well does anyone try to pretend doom scrolling make you smarter? I think platform companies have been successfully been turning people into morons for 20 years and now you don’t need to try even read a single news article or a blog post to learn how to solve a simple problem we are paving our way into intellectual (and literal) new dark ages.
Perhaps we go back to feudal society when climate change crumbles the civilisation, world economy and democracy. It didn’t really matter that French and Spanish kings where often literal morons, when you had few talented monks, bankers and scribes doing the brain-thing, the feudal lords had ruthlessness to take what they wanted and the people were illiterate superstitious folk who hardly ever left the village they were born in.
Doom scrolling is mindless entertainment, probably similar to the change from books to TV. It's not analogous to the loss of cognitive abilities we see with AI.
You will have hard time to prove it isn’t. We have plenty of studies showing how bad platform providers are for you in every which way: focus, learning ability, reading level etc. And no, it is in no way like TV (although TV today tries its hardest to resemble social media content).
Point is, the technology itself is not harmful nor innocuous in itself, it is how we let these shitty corporations to guide and control how we use technology. AI is a useful tool, so is chat application with a friend list. So is a hammer. You don’t need to use any of them to smash your face in.
> It didn’t really matter that French and Spanish kings where often literal morons, when you had few talented monks, bankers and scribes doing the brain-thing, the feudal lords had ruthlessness to take what they wanted and the people were illiterate superstitious folk who hardly ever left the village they were born in.
This resembles my country.
I'd love to see the cite for people experiencing cognitive atrophy and measured IQ loss from using AI.
I guess we’ll have to see how this affects the results in 2026’s global IQ census.
I fear there is some truth in it.
“Better for you if you take me off”
Prompt an AI with:
And you will get at least one recipe which surely is tasty.Ask a person the same question and you might be told:
This is the difference between a statistically probable response and understanding.