Ryan Lopopolo(@_lopopolo):And that generalizes to every auto-rater, every judge, every rubric, every eval, and every researcher as well. These misalignments compound over time. The models do not have a fear of future regret. Long-term coherence through agentic work product is very unsolved.

Ryan Lopopolo

@_lopopolo

Principal Engineer, Agentic GCP @googlecloud • prev @OpenAI • Building the future of work. Harness engineering. As an agent influencer, • opinions mine

加入 January 2022

1.2K 正在關注 10.6K 粉絲

Ryan Lopopolo@_lopopolo

2026.09.13 03:08

And that generalizes to every auto-rater, every judge, every rubric, every eval, and every researcher as well. These misalignments compound over time. The models do not have a fear of future regret. Long-term coherence through agentic work product is very unsolved.