Comment by InsidiousWombat
8 hours ago
Why would it be controversial, it's widely observable phenomena.
Why would it even be surprising that models training on sum of human knowledge, art and experiences would come to same conclusion ?
It's just hard for most to face the flaws most of us carry. Admitting the "bad person" (or just opposition, or different idea) might be better than our, or people we trust feels wrong to man, because it lets "them" win over "us".
It isn’t really controversial, of course, but it’s interestingly subtly sycophantic and self-promoting? You can see why RLHF might lead to such answers.