I think it's worth mentioning specifically that this approach causes a tension between following Anthropic's instruction and the model's "core value" of honesty.If the user asks it for the…
Posters: repligate. repligate
5 likes recorded by the source corpus.
Topics: AI & LLMs
View Post: https://x.com/repligate/status/1830732862657634372