Introdução à IA #2: Agentes e Racionalidade

Introduction to Agents and Rationality in AI

Understanding Agents

• The concept of an agent is introduced as a being capable of perceiving its environment through sensors and acting upon it via actuators.

• Human agents use sensory organs (eyes, ears) for perception and body parts (hands, legs) for action; robotic agents utilize cameras and motors.

• Software agents perceive inputs from keyboards or network packets, with outputs displayed on screens or sent over networks.

Agent Functionality

• The mapping of perceptions to actions is defined as the agent function, which mathematically describes the agent's behavior.

• An example using a simple vacuum cleaner illustrates how an agent perceives its environment (rooms A and B), determining whether they are clean or dirty.

Decision-Making Process

• The vacuum cleaner can take specific actions based on its location and the cleanliness of the room: moving left/right or cleaning.

• A partial table representing the vacuum cleaner's agent function shows how different perceptions lead to corresponding actions.

Programming Agent Behavior

Implementation of Agent Functions

• The programming aspect involves creating functions that respond to perceptions by defining conditions for actions based on room status.

Concepts of Rationality

• To design intelligent agents, understanding rationality, omniscience, and autonomy is crucial; omniscience implies complete knowledge of the environment.

Limitations and Practical Applications

Realistic Expectations from Agents

• An omniscient agent would maximize real-world performance but is impractical due to unpredictable events beyond control.

Rational Agents in Practice

• Rational agents aim to maximize expected performance based on current perceptions while learning from their environment for improved decision-making.