Re:GPT trainer says he's traumatized from the RLHF work
by wilg on Friday May 19, 2023 @04:43PM
(#36006638)
It’s a huge unsolved problem that content moderation of any kind requires humans looking at horrible content. Not sure what you can do beyond provide lots of mental health services or try to replace this whole thing with AI.
[ Parent
| on HN ]
Re:GPT trainer says he's traumatized from the RLHF work
by ok_dad on Friday May 19, 2023 @05:17PM
(#36006966)
I don’t want this, but an easy answer is to eliminate anonymity on those forums you want to be less-toxic.
A hard answer is to maintain a community properly to ensure self-policing with good backup moderation, like they do here at HN. There are only two (?) moderators but the culture here is very well maintained to self police threads.