Definition

A type of AI training where a system learns by doing something, seeing what happens, and trying to do better next time. It gets a reward for good choices and a penalty for bad ones—just like training a pet.

Example

“Reinforcement learning helps robots learn how to walk by rewarding them when they stay balanced.”

How It’s Used in AI

Used in robotics, self-driving cars, game playing (like AlphaGo), and automated stock trading. It’s great when an AI needs to learn by interacting with its environment over time.

Pro Tip

Reinforcement learning is powerful—but it takes time and lots of tries. It's best used when the AI can practice safely and often.