108. Last Week In AI — 2021: The (full) year in review
Towards Data Science
Is It a Good Policy to Keep Options Open?
This paper really struck me as the optimal policies sound a lot like samford underground. Wo are trying to open doors and try to be in a state where they have as many doors open as possible. But i think the coviat is like, does tot mean they willand i think when it is optimal, will select certain paths eventually. Soi thought it was interesting for that reason. I think this paper allows us to look at what are the real assumptions that are being baked into the safety argument,. And whateve ways that theory needs to approach more closely? It's exciting because its kind of like an attempt to bring that theory into something closer to practice.
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
