Effective Altruism: Ten Global Problems – 80000 Hours cover image

Four: Brian Christian on artificial intelligence

Effective Altruism: Ten Global Problems – 80000 Hours

00:00

How Can Reinforcement Learning Go Off the Rails?

Reinforcement learning is about how you develop a set of behaviors. Stuart russell, for example, who we've interviewed on this show, thinksthat we kind of need to do away with reinforcement learning for sufficiently advanced agents. So there's kind of a double optimization problem going on. You have some behaviour in mind, you need to create a reward which will in turn incentiviz that behavior.

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