Chapter 2: Intelligent Agents

2.1 What is an Agent?

An agent is anything that perceives its environment through sensors and acts upon that environment through actuators.

Formal Definition:
An agent is an autonomous entity that uses sensors to observe its surroundings (percepts), and actuators to act in that environment.

2.2 Agent-Environment Loop

Here’s how an agent works:

[ Environment ]
      ↓ (percept)
+-----------------+
|     Sensor      |
+-----------------+
      ↓
+-----------------+
|   Agent Logic   |
| (decision-making) |
+-----------------+
      ↓
+-----------------+
|    Actuator     |
+-----------------+
      ↓ (action)
[ Environment changes ]

The agent continuously:

Observes the environment (percepts)
Thinks (based on built-in rules or learning)
Acts to change the environment

2.3 What is a Rational Agent?

A rational agent is one that does the right thing — i.e., it acts to maximize its performance measure based on the percept history and its knowledge.

2.4 Performance Measures

Agent	Performance Measure
Chess-playing AI	Win the game
Delivery Robot	Deliver to correct location efficiently
Email Classifier	Correct spam detection rate
Smart Vacuum	Clean the floor quickly and thoroughly

2.5 The PEAS Framework

PEAS = Performance measure, Environment, Actuators, Sensors. This framework helps you design or analyze an AI agent.

Example: Autonomous Taxi

Component	Description
Performance	Safe driving, fast delivery, obey traffic laws
Environment	Roads, traffic, passengers, pedestrians
Actuators	Steering, accelerator, brake, display, horn
Sensors	GPS, camera, radar, LiDAR, speedometer

2.6 Agent Types (with Diagrams)

There are five standard types of agents, progressing in complexity:

Simple Reflex Agent: Acts only on current percept using IF-THEN rules.
Model-Based Reflex Agent: Uses internal state (memory) to keep track of unseen parts of the world.
Goal-Based Agent: Makes decisions to achieve a goal.
Utility-Based Agent: Chooses the most desirable outcome when multiple goals exist.
Learning Agent: Improves its performance over time through experience.

2.7 Summary Table: Agent Types

Type	Memory	Goals	Learning	Example
Simple Reflex	❌	❌	❌	Basic vacuum cleaner
Model-Based Reflex	✅	❌	❌	Wumpus agent
Goal-Based	✅	✅	❌	Maze-solving robot
Utility-Based	✅	✅	❌	Route optimizer
Learning Agent	✅	✅	✅	AI playing chess, spam filter

2.8 Common Exam Questions

Define an intelligent agent with two examples.
What are the differences between reflex and goal-based agents?
Explain the PEAS framework using an interactive tutor.
Draw and explain the architecture of a learning agent.
What makes a rational agent different from a rule-based agent?