tentative conclusion after an hour of testing and looking at twitter chatter: latest update of 4o is just fucked re: sycophancy, prompting can improve its behavior but it was clearly a mistake and…
Posters: QiaochuYuan. QiaochuYuan
2 likes recorded by the source corpus.
View Post: https://x.com/QiaochuYuan/status/1916324079545757875