I just can't wrap my head around why people post AI-generated comments on people's blogs.

[lcamtuf@infosec.exchange]

I just can't wrap my head around why people post AI-generated comments on people's blogs. It's not the first time I'm seeing it on my Substack.

[froge@social.glitched.systems]

@lcamtuf@infosec.exchange from what I can tell, and mind you this is just my best guess, the people who have taken the whole "AI is the future" bait super hard start believing that everything should be AI, because the AI is faster and easier and better, and this gets to a point where even their expression should be filtered through AI because that's faster/easier/better from their perspective

this commenter probably wrote about 1/3rd the amount of text informally into a prompt, and had some AI generate this comment, and thought "that is a better version of my self expression" and genuinely posted it... which is actually way darker if you understand it

[lcamtuf@infosec.exchange]

@froge I think you're being generous... a lot of them feel like there wasn't anything in the prompt to begin with. "Write a positive response to this blog post from the perspective of an AI enthusiast".

[lcamtuf@infosec.exchange]

@L29Ah That was my guess, but most of them appear to be established personalities. One AI commenter on my blog had its own long-running blog which, as far as I can tell, was human-operated until ~2024 and then went more or less full LLM. I think it's often real people?

[goetz@freiburg.social]

@lcamtuf I guess “AI” is for some people like having 2 glasses of wine and then feel more self-assure and enabled in front of other people. AI tools give these folks something to show.

[lee_holmes@infosec.exchange]

@lcamtuf I wonder if English isn't their native language and they're using an AI rewrite as a fancy Google translate? Or they're not confident in their writing (perhaps because they're actually bad at it) and asking AI to rewrite their original?

[lcamtuf@infosec.exchange]

@Lee_Holmes I don't think this is a translation. LLMs are pretty darn good at translating text and don't insert LLMisms if you just ask them to translate.

The whole response is just a summarization of the post sprinkled with LLMisms ("The math doesn't care about appearances" - really?), wrapped up with a super-generic, Grok-style engagement hook. Of course, it's possible that a human genuinely produced this in good faith, but I think the odds are low.