Definition: reinforcement learning

A method for training AI models by providing a reward (higher score) for actions that achieve a goal or a penalty (lower score) for actions that do not. Reinforcement learning from human feedback (RLHF) requires an AI engineer to evaluate the actions (see RLHF).