BPSY361: Introduction to AI Quiz

Course Information

• Course Title: BPSY361: Artificial Intelligence (AI)
• Institution: CHRIST (Deemed to be University)
• Location: Bangalore, India
• Mission: Holistic development to contribute effectively
• Vision: Excellence and Service
• Core Values: Faith in God, Moral Uprightness, Love of Fellow Beings

Unit 1: Introduction

• Introduction to AI: Covers basic concepts, intelligent agents, agents and environments, good behavior, nature of the environments, structure of agents, problem-solving, and examples of problems

Artificial Intelligence (AI)

• Artificial: Created by humans, in imitation of natural things

What is Artificial Intelligence in AI?

• AI uses computers and machines to mimic human problem-solving and decision-making abilities.

What is Intelligence?

• Ability to acquire and apply knowledge and skills
• AI is a method to make a computer, computer-controlled robot, or software think intelligently like a human mind
• Psychologists define intelligence as the ability to learn, recognize and solve problems

Intelligence (Brain and its Hemispheres)

• Stronger connection between brain hemispheres
• Einstein (1905): Had more extensive connections between certain parts of his cerebral hemispheres compared to younger and older control group brains.

Introduction to AI (Field, Science, and Data)

• AI is a field that combines computer science and robust datasets to enable problem-solving

A Brief History of Artificial Intelligence

• 1956: John McCarthy coined the term "artificial intelligence" and held the first AI conference.
• 1969: Shakey, the first general-purpose mobile robot.
• 1997: IBM's Deep Blue defeated a world chess champion.
• 2002: First commercially successful robotic vacuum cleaner.
• 2005-2019: Advancements in speech recognition, robotic process automation (RPA), and smart homes.
• 2020: Baidu's LinearFold AI algorithm developed a Covid-19 vaccine faster than previously possible

Intelligent Agents

• Agent: A computer program or system designed to perceive its environment.
• Make decisions and take actions to achieve specific goals.
• Operate autonomously.
• Diagram shows sensors, percepts, environment, actions, and effectors.

Intelligent Agents: Agents and Environments

• Rational Agents: Logic or Reason based systems called intelligent
• Agent: Anything capable of perceiving the environment with sensors and acting upon it with actuators.
• Agent behavior is defined by an agent function that maps percept sequences to actions.
• Diagram shows Agents interacting with environment through sensors and actuators

Agents and Environments (Agent Function)

• Agent function: Abstract mathematical description of an agent, contrasting with the concrete implementation (agent program), which runs within a physical system

Example: Vacuum-Cleaner World

• Diagram shows a vacuum-cleaner world with two locations (A and B).
• Table displays percept sequences and corresponding actions.

What is Agent and Environment?

• An agent is something that senses and acts upon its environment.
• Humans have sensory organs (eyes, ears, etc.) and effectors.
• Robotic agents use sensors like cameras, IR, and effectors like motors.
• Software agents have coded programs and actions.

Agent Terminology

• Performance Measure: Criteria for evaluating an agent's success.
• Behavior: Action taken by the agent based on percepts.
• Percept: Input received by the agent at a specific time.
• Percept Sequence: The history of all external inputs the agent has received up to a given point.
• Agent Function: Mapping from percept sequence to action.

Good Behavior: The Concept of Rationality

• A rational agent does the right thing conceptually, ensuring correctness of every entry in its agent function table.
• The agent's rationality depends on the performance measure, prior environment knowledge, actions, and current percept sequence.

Definition of a Rational Agent

• For each possible percept sequence, a rational agent selects the action it expects to maximize its performance measure given the evidence provided by the percept sequence and built-in knowledge.

Rules for AI agents

• Rule 1: An AI agent must have the ability to perceive the environment.
• Rule 2: Observation must be used for decision-making.
• Rule 3: Decisions must result in an action.
• Rule 4: Actions must obey a rational course of action

Example (Performance Measure)

• The performance measure awards one point for each clean square at each time step during 1,000 steps.
• Environment geography is known but dirt distribution and initial agent location are not.
• Actions are Left, Right, and Suck

Omniscience, Learning, and Autonomy

• Omniscience: An agent knows the outcome of its actions, learns through information gathering, and adapts with pre-existing knowledge.
• Autonomy: Agent learns from its environment to compensate for partial or incomplete knowledge

The Nature of Environments

• Task Environments are essentially problems where solutions are rational agents.
• PEAS is a model covering Performance, Environment, Actuators, and Sensors

Properties of Task Environments

• Fully Observable vs. Partially Observable (environment and agent)
• Deterministic vs. Stochastic (environment's state)
• Episodic vs. Sequential (temporal dependency between actions)
• Dynamic vs. Static (if environment continuously changes)
• Discrete vs. Continuous (number of possible actions or steps)
• Known vs. Unknown (agent’s knowledge of environment)

Structure of Agents

• Agent = Architecture + Agent Program
• Architecture is machinery. Sensors and Actuators.
• Agent Program is an implementation of an agent function.
• Agent Function is a map from percept sequence (current agent's perception) to action.

Structure of Agents (Agent Types)

• Simple reflex agents
• Model-based reflex agents
• Goal-based agents
• Utility-based agents
• Learning Agent

Simple Reflex Agents

• Simplest type. Directly responds to percepts.
• Uses condition-action rules (if condition then action).
• Example: Car braking given in-front car braking.

Model-based Reflex Agents (World Tracking)

• Handles partial observability by tracking the world.
• Combines current percept and previous internal state to update world knowledge.

Goal-Based Agents (Specific Desirable States)

• Agents have descriptions of desirable states (goals).
• Action selection depends on goal states.

Utility-Based Agents (Quality of Action)

• Agents choose actions based on utility (quality) rather than simply reaching a goal.
• Selects actions that lead to more reliable and more desirable outcomes

Learning Agent (Adaptable Agent)

• Learns from past experiences and adapts automatically.
• Component parts include: Learning Element, Critic, Performance Element, and Problem Generator

Problem Solving Agents

• Problem-solving agents determine sequences of actions toward a desirable state.
• Search is a method to find such a sequence, involving a systematic exploration of alternative actions.
• Problem definition is a detailed specification of acceptable solutions and inputs.
• Problem analysis, Knowledge Representation, and problem solving techniques are parts of the process.

Components of Problem-solving (Phase descriptions)

• Goal Formulation: Agent formulates specific steps to reach target/goal.
• Problem Formulation: Fundamental step in the problem-solving process, it details what action to take to reach the goal.
• Search: Finding the optimal sequence of actions that will achieve the goal. Simulates multiple potentially unsuccessful sequences of actions.
• Execution: Executing a sequence of actions that was identified through search

Components of a problem (Formal definitions for the process)

• Initial State: Agent's starting state or initial step toward the goal.
• Actions: Possible actions an agent can take (based on the given state).
• Transition Model: Describes the result of an action in a given state.
• Goal Test: Checks if the state matches the goal.
• Path Cost: An agent's numerical cost of leading to the goal (usually the goal's lowest possible cost to determine optimal solutions).

Example Problems (Problem types)

• Standardized/Toy Problems: Used to demonstrate problem-solving techniques (e.g., puzzles).
• Real-world problems: Require solutions for scenarios in the real world (e.g., route-finding, touring problems, Traveling Salesman Problem).

Example problems (Specific Problem types)

• 8-puzzle: A puzzle where tiles need to be arranged in a specific order.
• 8-queens problem: A chess puzzle where 8 queens need to be placed on a board so that no two queens attack each other.
• Route-finding problem: To find the shortest route between locations.
• Touring problems: To visit all locations or cities once and return to the starting location

The Traveling Salesman problem

• The salesman must travel to all cities once before returning home.
• Objective: Minimize the total distance traveled