how does chatgpt work?

Explore/Exploit Tradeoff in Reinforcement Learning