OpenAI's "Study Mode" and the risks of flattery

Re:OpenAI's "Study Mode" and the risks of flattery by furyofantares on Saturday August 02, 2025 @11:07AM (#44768227)

If you don't mind me asking - was this a very long single chat or multiple chats?

[ Parent | on HN ]

Re:OpenAI's "Study Mode" and the risks of flattery by neom on Saturday August 02, 2025 @11:37AM (#44768497)

Multiple chats, and actually at times with multiple models, but the core ideas being driven and reinforced by o3 (sycophant mode I suspect) - looking back on those few days, it's a bit manic... :\ and if I think about why I feel it was related to the positive reinforcement.

[ Parent | on HN ]

Re:OpenAI's "Study Mode" and the risks of flattery by furyofantares on Saturday August 02, 2025 @01:09PM (#44769316)

Thanks for posting about it.

[ Parent | on HN ]