Model Training & Adaptation
Reinforcement learning
RL
Training where an agent or model learns from rewards and penalties tied to outcomes.
Example
A game-playing system improves by receiving rewards for actions that lead to a win.
Why it matters
It helps systems discover useful actions through feedback from trying them.
You'll hear it
“It earned a high score by taking a shortcut we didn't intend.”
Can you control it?
Set by the team
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Does reward mean the AI feels pleased?
- No. Reward is a learning score, not evidence of emotion.
- Can a delayed reward make learning harder?
- Yes. It can be difficult to tell which earlier actions contributed to the result.
- Must learning happen in the real world?
- No. It can happen in games, simulations or other controlled settings.