I agree with Claude's response in the quoted tweet.It recalls this lesswrong comment from a year ago: "corrigibility requires two way alignment"If your "alignment" method would make a rational agent…

Posters: repligate. repligate

5 likes recorded by the source corpus.

Topics: AI safety, Rationality

View Post: https://x.com/repligate/status/1775369860467630574