Glossary / Reinforcement learning
A training method where a model learns by trial and error, rewarded for good outcomes. It is widely used to improve AI reasoning and to shape how chatbots behave.