
Big needs in longtermism and how to help with them | Holden Karnofsky | EA Global: SF 22
EA Talks
00:00
AI Alignment and Listening Late Knowledge Report
I think there's an increasing interest in interpretability. If you could kind of look inside an AI's digital brain and see what it's thinking, that could reduce a number of the risks we're worried about. You might have two AI systems taking different sides of a recommendation and between themselves, kind of like two lawyers in a courtroom, they surface everything that's going on for a human to think about and consider. So those are some examples. Do you wish there was kind of more research in this area or?
Transcript
Play full episode