The Inside View cover image

Tony Wang on Beating Superhuman Go AIs with Advesarial Policies

The Inside View

00:00

The Intent Alignment Problem

I think about the Paul Christiano definition I think what he was posted like six years ago about like a system is aligned to the person operating it. So if I tell my robots please like like cook me some meal. It's all tried to do what I kind of mean when I said in my cell it might not be like super typical that doing it. And then I think that's a very good definition but then then most people with different goals like opening eye under user is going to wear to like talk about this definition, even if you're only one or two users. Yeah. Maybe my main concern with that definition is once you have two people. Then if you feel like multiple people

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app