by Albert Major

Introduction to Artificial Intelligence

Chapter 3 - Solving problems by searching

Problem-Solving Agents

What it is: An AI that plans ahead by considering sequences of actions to reach a goal

• Atomic representations
  • Definition**:** States of the world considered as wholes, without internal structure - (smallest building blocks)
• There are two types of search algorithms:
  • Informed: In which an agent can estimate how far it is from its goal
  • Uninformed: In which in which an agent can not estimate how far it is from its goal

Agents (usually) follow a four stage process:

1. Goal formulation:
   • The process of identifying and clearly defining the objective or desired outcome of a problem-solving task
2. Search
   • The computational process of finding a sequence of actions to reach a goal
3. Problem formulation:
   • The process of defining the initial state, possible actions, and goal state of a problem in a way that can be understood and solved by a problem-solving agent
4. Execution:
   • The process of carrying out the actions in the solution plan to achieve the goal in the real world.

• In a fully observable, deterministic, known environment, the solution to any problem is a fixed sequence of actions
• So, there are two types of systems:
  • Open-loop: A system that carries out a predetermined sequence of actions without considering feedback or changes in the environment.
  • Closed-loop: A system that monitors its actions and the environment, adjusting its behavior based on feedback to ensure it reaches the goal.

Types of Environments:

• Episodic: Each action/decision is independent of previous ones
• Single agent: Only one AI is making decisions in the environment
• Fully observable: The AI can see all relevant information about the environment