[zahlman]

For what it's worth: having seen that someone else suspected ChatGPT usage, and reading it again, I can understand what sorts of heuristics it might have tripped. But on overall intuition, I didn't get that impression on a first read.

[fc417fc802]

If LLMs had the necessary theory of mind to model context well enough to write the comment in question the world would be a very different place. Writing style and fitting in to the context are entirely different things.

LLM responses tend to give me the vibe of a disengaged office worker who isn't emotionally present and lacks a proper mental model of the topic. It's been shocking to me to witness how frequently people around me IRL fail to notice such writing.

[zahlman]

> Writing style and fitting in to the context are entirely different things.

Absolutely — which goes a long way towards how I make the judgment myself, now that you say it. Although it's still hard to be precise about.

(One positive impact that LLMs indirectly had on my life is that they motivated me to look up ways to type em-dashes more easily. I now have my caps lock key mapped as a compose key, and a custom compose-key mapping for it.)

> It's been shocking to me to witness how frequently people around me IRL fail to notice such writing.

To notice, they'd have to themselves be engaged, I suppose. Or at least have some idea about the subject matter.