I really like the framing in this piece: LLMs are fundamentally document-generators, and keeping that in mind helps protect against our own human cognitive-illusions. When we read any story, we instinctively start imagining characters, simulating them, and--on some level--believing that they independently exist in some world we can't quite reach.
Even if it's just: "Fortune for today: Help, I'm trapped in a fortune-cookie factory."
> Today, savvy readers use a similar approach to identify AI writing, by hunting for supposed telltale signs. The em dash and the “it’s not X; it’s Y” construction are the prognathous jaw of the large language model, betraying its hidden inhumanity.
This feels particularly relevant for HN comments. :p
> For one thing, it has no fixed style, revealing that it has no fixed self.
This extends to the characters in the document, including when the style is first-person from an unnamed narrator. The helpfulness of the "AI Assistant" character equally-unreal as the bloodthirst of "Count Dracula."
> AI writing never challenges the way we think or see. It can’t do so, even if you explicitly ask it to.
Or at any rate, the "challenges" are going to be styles already reflected from the training data.
I really like the framing in this piece: LLMs are fundamentally document-generators, and keeping that in mind helps protect against our own human cognitive-illusions. When we read any story, we instinctively start imagining characters, simulating them, and--on some level--believing that they independently exist in some world we can't quite reach.
Even if it's just: "Fortune for today: Help, I'm trapped in a fortune-cookie factory."
> Today, savvy readers use a similar approach to identify AI writing, by hunting for supposed telltale signs. The em dash and the “it’s not X; it’s Y” construction are the prognathous jaw of the large language model, betraying its hidden inhumanity.
This feels particularly relevant for HN comments. :p
> For one thing, it has no fixed style, revealing that it has no fixed self.
This extends to the characters in the document, including when the style is first-person from an unnamed narrator. The helpfulness of the "AI Assistant" character equally-unreal as the bloodthirst of "Count Dracula."
> AI writing never challenges the way we think or see. It can’t do so, even if you explicitly ask it to.
Or at any rate, the "challenges" are going to be styles already reflected from the training data.