Reinforcement learning

1988

Learn from reward rather than labels. TD learning 1988, Q-learning 1989, then deep RL and RLHF.

Also: RL