Episode 22: Archit Sharma, Stanford, on unsupervised and autonomous reinforcement learning
Generally Intelligent
00:00
Do You Want to Learn How to Get a PhD for Repetition Learning?
Do we really want to learn optimal behaviors? Or do we want to learn how to just finish a task? And arguably, like often the cases that I just want something done, maybe even if it's that optimal, like it's good enough is fine. So encoding that inner optimization for ill robots would be very, very interesting as well. Especially for a generalist region, so we are thinking about that.
Transcript
Play full episode
