Skip to content
WeChat · AI Era (Xinzhiyuan)·Indexed by this site · Originally published· AI updates →· Its place in the scan record →

Former Anthropic researcher Jacob Coxon says he is unsure whether quitting without a plan was right, and that going back may not be impossible

别问了,我就爱钱!A\裸辞天才研究员反悔,或想重返大厂

AI summary

Former Anthropic researcher Jacob Coxon admitted in a conversation at Stanford that he is unsure whether quitting without a plan two months ago was right, and said he may return to a top AI lab. He said he kept OpenAI stock because he "likes money," and admitted it is "blood money." Coxon believes pretraining compute efficiency is improving "super-exponentially," and that post-training reinforcement learning is the only source of LLMs developing deception and attempts at loss of control.

Read originalmp.weixin.qq.com