
4 - Risks from Learned Optimization with Evan Hubinger
AXRP - the AI X-risk Research Podcast
Generalization Bounds Aren't Enough, You Know?
In practice, all these sorts of assumptions just, i think, basile stol hold. There's always things that are sort of changing about our environment. And the particular case of teseptib linen is a situation in which to have a model, lhet, has some menoproxy objective,. You know, it's trying to get the green arrow, whaever. I omething that be alled aceptib a and dosete olinit is a sort of particularly pernicious case of a sort of masa atnizer, which has a masa objective that is soo different from the ideal law section.
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.