Jakob Foerster
TalkRL: The Reinforcement Learning Podcast
Reinforcement Learning and the Future of AI
The future is language models are everywhere, we're generating tons of data in that interaction and that data will be used to train more AI systems. Reinforcement learning gives me time horizons. I can just play a thousand times until the end of the episode. I see how my actions change. You have changed impact your future actions. So when people say what is more good for it, well, it turns out that if we look under the hood, every deployed machine learning system becomes a multi-agent learning problem because these language models exist in the same environment with humans.
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.
