Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

The sycophancy is also a concern, it’s not really an impartial teacher, all its training is to suck up and maximize engagement rather than learning. The incentives are wrong.


> all its training is to suck up and maximize engagement rather than learning

That's speculative, isn't it


No, that's a direct consequence (intended or not) of how RLHF works.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: