Hacker News Reader

Comment by circuit10 | Hacker News Reader

I think it matters because AI can hallucinate so it makes me doubt the truthfulness

It's also somewhat dishonest to present unedited AI-written content as your own work

That's a fair point. For factual content, AI output should be verified rather than trusted blindly because hallucinations are real. I see AI more as something to challenge and verify my thinking than as something to make decisions for me.

People also say things that are untrue? So I guess it boils down to reputation and attribution. I kinda feel like there is a wierd knee jerk to pretend like “ before AI, if it was written, you could trust it” which absolutely was never true.

Thing is that pre-AI even for that you need to spend time and effort. Now those deterrent barriers are not anymore so that kind of content can be produced massively at industrial scale.

Automated (semi)intentional lying on internet predates LLMs

Generally, people have to make an active decision to spread false information, or be misinformed themselves in the first place, and that's not actually particularly common, whereas for LLMs it's just the default due to how they work and it's difficult to filter out due to how they present everything with such confidence and authority

"Somewhat dishonest"? It's plagiarism. Just because you claim the output of a machine as your own instead of the writing of another human as your own doesn't somehow make it yours and not plagiarism.

Reply on news.ycombinator.com