Artificial Intelligence: Agents, Environments, and Search Methods

Core Concepts of Artificial Intelligence

Defining Key AI Components

• Artificial Intelligence (AI): A study to perceive, reason, and act. Systems that think like humans, trying to understand how the human mind works. Systems based on rules.
• Intelligent Agent: An entity capable of perceiving its environment, processing these perceptions, and responding or acting in that environment rationally and appropriately.
• Prolog: A functional programming language based on logic.
• Rational Agent: Agent = Architecture + Program. The program depends on the environment in which it operates.

Types of Intelligent Agents and Programs

The agent program dictates how the agent chooses its actions:

1. Simple Reflex Agent

   Actions depend solely on the current perception and predefined rules applied.

   Program Structure: state <- interpret_inputs; rules <- match_rules; actions <- actuation_rules

2. Reflex Agent with Internal State

   Uses an internal state to track the world history.

   Program Structure: states <- modify_states; rules <- match_rules; action <- actuation_rules; State <- change_state

3. Goal-Based Agent

   Maintains control of the world based on specific objectives.

   Example: A taxi driver must drive to a destination.

   Tools: Search and planning algorithms.

   Action: target_algorithm (state, perception)

4. Utility-Based Agent

   Generates all possible states and chooses the action that maximizes the utility function (the most appropriate assessment).

Characteristics of AI Environments

Environments are classified based on five key properties:

1. Accessible vs. Inaccessible

   The agent can see all necessary information / The agent does not have all the information (partial observability).

2. Deterministic vs. Nondeterministic

   The best move can be determined / Cannot choose the next state definitively from the current one.

3. Episodic vs. Non-episodic (Sequential)

   Actions in one episode do not affect the next / Thinking before acting (current action affects future states).

4. Discrete vs. Continuous

   Discrete quantity of moves or states / Continuous variables, difficult to quantify.

5. Static vs. Dynamic

   The environment remains unchanged while the agent is thinking / The environment changes over time.

AI Search Algorithms and Strategies

Uninformed and Informed Search Methods

1. Uninformed (Blind) Search

   There is no information regarding the number of steps or the cost of the path from the current state to the final state. Includes Breadth-First and Depth-First search.

2. Heuristic Search (Informed Search)

   Uses a heuristic function h(n) indicating the estimated cost of the search from the current state (n) to the final state.

3. Best-First Search

   The node that receives the best evaluation is expanded first. This function takes into account all nodes seen so far to determine the best path.

Specific Search Algorithms

4. Breadth-First Search (BFS)

   Each node is evaluated completely at a certain level before moving on to the next level. It is optimal (finds the shortest solution).

   How it works: Searches the entire graph or sequence without considering the goal until it is found. It is an uninformed search (non-heuristic). All child nodes obtained by expanding a node are added to a FIFO (First-In, First-Out) queue.

5. Depth-First Search (DFS)

   An algorithm that allows recursively traversing all nodes of a graph or tree in an orderly but non-uniform manner.

   Its operation involves expanding every node located recurrently in a specific path. When there are no more nodes, it backtracks and repeats the process with each unprocessed branch node.