AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
How to Get an Agent to Not Forget and Continuously Learn?
A lot of the work in my career has been focused on the question of what we could call broadly exploration. So how do you get an agent to intelligently explore a new environment and continuously learn forever? And it turns out that actually is way better ultimately for solving problems in many cases than just trying to directly encourage the robot to solve the problem. It's really cool to see all of these threads kind of emerging and connecting throughout all this work.