Rational Agent
Is an agent that [see page 24, acts] to achieve the best outcome or when there is uncertainty the best expected outcome. That is an agent that behaves rationally.
Rational agents use a performance measure which given a sequence of environment
states, evaluates how good they are. A sequence of states for a cleaning robot may
observe the robot is staying in the same clean square forever and give it a low score
to indicate it should explore a bit.
Note: performance is measured in terms of the environment states, not the agent states. If the agent could grade itself based on its own representation of the world, it could delude itself into thinking it's perfect.
As a general rule, it is better to design performance measures according to what one actually wants in the environment, rather than according to how one thinks the agent should behave.
The rationality of such an agent is dependent on:
- The performance measure (criterion of success)
- The agents prior knowledge of the environment
- Actions available to the agent