Types of Agents

This section explores the four main architectural designs for intelligent agents, ranging from simple to complex.

---

1. Simple Reflex Agents

"Do X if Y happens right now."

• Concept: These agents select actions based only on the current percept, ignoring the rest of the percept history.
• Requirement: The environment must be fully observable (the agent needs to see the relevant part of the world to make the right decision).
• Logic: Condition-Action Rule (If Condition then Action).

Example: Thermostat

• Percept: Current Temperature = 15°C
• Rule: IF temperature < 20°C THEN Turn Heater ON
• Action: Turn Heater ON
• (It doesn't care what the temperature was 5 minutes ago, only right now).

---

2. Model-Based Reflex Agents

"Do X based on what I see now AND what I remember."

• Concept: These agents maintain an internal state (history) to keep track of aspects of the world that are not currently visible.
• Why: Necessary for partially observable environments.
• Key Component: A "Model" of the world (knowledge about how the world works and evolves).

Example: Autonomous Car Changing Lanes

• Percept: The car in front is slow.
• Internal State: Remembered that a fast car was approaching in the left lane 2 seconds ago (even if it's in the blind spot now).
• Action: Wait before changing lanes.

---

3. Goal-Based Agents

"Do X because it helps me get to Y."

• Concept: The agent acts to achieve a specific Goal.
• Difference: Knowing the state of the world isn't enough; the agent needs to know what it's trying to achieve.
• Capability: Search and Planning. The agent considers the consequences of its actions.

Example: GPS Navigation (Google Maps)

• Current State: At Home.
• Goal: Reach the Airport.
• Decision: The agent (app) calculates multiple routes and chooses the turn that gets it closer to the airport, not just any random turn.

---

4. Utility-Based Agents

"Do X because it's the BEST way to get to Y."

• Concept: Goals are not enough (there might be many ways to reach the airport). We want the best way.
• Utility Function: Maps a state to a real number (score) representing how "happy" or "successful" the agent is.
• Focus: Maximizing Utility (Efficiency, Speed, Safety, Cost).

Example: Taxi Routing

• Goal: Reach destination.
• Utility Factors: Time taken, Fuel cost, Passenger comfort.
• Decision: Choose the route that is 5 minutes faster (Higher Utility) even if it's slightly longer in distance.

---

5. Learning Agents

"How can I improve my behavior based on experience?"

Learning allows the agent to operate in initially unknown environments and to become more competent than its initial knowledge alone might allow.

Four Conceptual Components:

1. Learning Element: Responsible for making improvements.
2. Performance Element: Responsible for selecting external actions (this is what we previously considered the entire agent).
3. Critic: Observes the world and provides feedback to the learning element on how the agent is doing relative to a fixed performance standard.
4. Problem Generator: Suggests actions that will lead to new and informative experiences (exploration).

Example: Learning to Play a Game

• Performance Element: Playing a move.
• Critic: Noting that the move led to losing a piece (negative feedback).
• Learning Element: Updating the strategy to avoid that move in the future.
• Problem Generator: Suggesting a new opening to see if it works better.

---

Summary of Agent Structures

Agent Type	Knowledge Used	Environment Requirement	Main Process
Simple Reflex	Condition-action rules	Fully observable	Match current percept to rule
Model-Based	Internal state + Model	Partially observable	Update state, then match rule
Goal-Based	Goals + Consequences	Predictable	Search and Planning
Utility-Based	Utility function (Happiness)	Multiple solutions	Maximize expected utility
Learning	Feedback + Exploration	Unknown	Improve over time

---