Reinforcement learning is the kind of machine learning in which an AI learns through experience with an environment, while feeling penalties or rewards based on feedback.
An agent in the case of reinforcement learning decides within an environment and modifies its actions according to outcomes received. This tries to maximize cumulative rewards over time.
Applications of RL have been widespread across games, robotics, and self-driving cars, which are learning through experience.
Unlike in supervised learning, wherein data labels are provided, in RL, one learns by trying and failing many times to uncover the best possible strategy for executing a task.