Probably, there's little reason it wouldn't. The profit (karma, attention) is too high for the cost (time, effort).

I was thinking the real profit was in "shaping the narrative". But yeah, I can see karma and attention being competitive.

I believe I've seen some comments that may have been LLM-generated. Typically they're top-level, short (1-3 sentences), somewhat on topic, but barely relevant to the conversations happening.

Like, for example, on a post of a blog post about a malicious package found in NPM is I'll sometimes see a post like; "NPM is a godsend for developers. Problems can occur, but it has helped developers move more quickly and has made sharing code a breeze."

I never report it since I'm not quite sure if it's generated by a program, or merely just a low-effort comment from a human. Which is tricky.

It's anyone's guess what will happen in the long run (or rather, the medium run) but for now the approach that seems to be working is for users to flag the comments that they guess are bot-generated. That's actually the approach we've always taken—it's just that the bots, as everyone knows, have taken a quantum leap recently.