Hacker News Reader

I agree with the rest of your comment, but it's framed as a retort but doesn't really work as one. The person your responding to already said "otherwise good answer" which excludes a wrong but "good-looking" one.

View on HN

Moderators generally aren't checking for correctness - who would have that sort of time?

I would expect the moderators to have some subject knowledge and often it doesn't take much to spot that an answer is blatantly fantastical as LLMs sometimes are. The finer points of some API might need actual checking, but a lot of things you'll know on sight.

Point is, you don't know which are wrong unless you do the work. The moderator has to do the actual work, but the LM (or rather, the user running an LM bot on his account) gets the credit.

They asked why it is a problem, and I tried to explain why it is in practice.

Reply on news.ycombinator.com