BPSY361: Introduction to Artificial Intelligence

Introduction to BPSY361: Artificial Intelligence (AI)

• CHRIST (Deemed to be University) offers BPSY361: Artificial Intelligence (AI).
• Mission: To nurture holistic development.
• Vision: Excellence and Service.
• Core Values: Faith in God, Moral Uprightness, Love of Fellow Beings.

Unit 1: Introduction

• Introduction to AI: Basic concepts, Intelligent Agents, agents and environments, good behavior, nature of environments, structure of agents, problem solving, and examples of problems.

Artificial Intelligence (AI)

• AI is created by humans, especially for imitation of natural things.

What is Artificial in AI?

• Artificial intelligence leverages computers and machines to mimic the problem-solving and decision-making capabilities of the human mind.

What is Intelligence?

• The ability to acquire and apply knowledge and skills.
• Artificial Intelligence (AI) is a method of making a computer, a computer-controlled robot, or software think intelligently like the human mind.

Intelligence: Key Insight

• Einstein had more extensive connections between certain parts of his cerebral hemispheres compared to both younger and older control group brains (1905 Miracle Year).

Introduction to AI (Updated info)

• AI is a field that combines computer science and robust datasets to facilitate problem-solving.
• A reference is provided: https://positivepsychology.com/artificial-intelligence-in-psychology/#apps

Foundations of AI: Philosophy

• The confirmation theory of Carnap and Hempel aimed to analyze the acquisition of knowledge from experience.
• Carnap's work (1928) outlined a computational procedure for extracting knowledge.
• This was potentially the first mind theory as a computational process.

Foundations of AI: Mathematics

• A foundational contribution to AI is the theory of probability.
• Gerolamo Cardano (1501–1576) initiated the concept of probability.
• Pascal's and Fermat's work provided methods for predicting gambling outcomes.
• Bernoulli, Laplace, and others further developed probability theory and statistical methods.
• Bayes' rule underpins modern approaches to uncertain reasoning in AI systems.

Foundations of AI: Economics

• Economics and operations research contributed significantly to the concept of rational agents.
• However, AI research initially diverged from economic approaches.
• Herbert Simon's (1916-2001) satisficing work, recognized with the Nobel Prize in 1978, proposed "good enough" decisions over optimal ones, providing a better model of human behaviour.
• The 1990s witnessed renewed interest in decision-theoretic techniques for agents.

Foundations of AI: Neuroscience

• Brain activity measurement started in 1929 with the EEG invention of Hans Berger.
• Functional magnetic resonance imaging (fMRI) and single-cell recording advancements offer detailed brain activity observation.
• These measurements relate to cognitive processes.

Foundations of AI: Psychology

• AI in psychology started at a 1956 MIT workshop.
• Contributors like George Miller, Noam Chomsky, and Allen Newell showed how computer models can solve memory, language, and logic problems.
• Psychologists increasingly view cognitive theories as computer programs that detail cognitive function.

Foundations of AI: Computer Engineering

• The first operational computer was the electromechanical Heath Robinson (1940).
• Colossus (1943) advanced to vacuum tube computers.
• Konrad Zuse's Z-3 (1941) and Plankalkul were pioneering programmable computers and languages.
• Atanasoff-Berry Computer (ABC) (1937-1942) was a precursor to modern computers.
• UNIVAC (1940-1942) and other projects built upon this technology.
• AI has transformed computer engineering through quantum computing advancements (optimization of algorithms, error mitigation, etc.).

Intelligent Agents

• An agent is a computer program or system designed to perceive its environment.
• It makes decisions and takes actions to achieve goals.
• The agent operates autonomously.

Intelligent Agents: Environments

• Systems that can reasonably be deemed intelligent.
• An agent is anything that perceives the environment through sensors and acts upon it through actuators.
• Agent behavior is described by a function that maps perceived sequences to actions.

Intelligent Agents: Agents and Environments

• Agent function is an abstract mathematical description.
• Agent program is the concrete implementation running within a physical system, such as a computer.
• Example: vacuum cleaner world with actions like "Right", "Suck", "Left" etc

Intelligent Agents: Properties

• Agent Terminology:
  • Performance Measure: Determines success criteria.
  • Behavior: Agent's action after perceptual inputs.
  • Percept: Agent's current perceptual input.
  • Percept Sequence: History of all perceived inputs till date.
  • Agent Function: Map that maps percept sequence to an action.
• Good Behavior (Rationality):
  • A rational agent does the right thing.
  • Rationality relies on performance measure, prior knowledge, actions, and perceived sequence.
• Definition of a Rational Agent:
  • A rational agent selects actions maximizing its performance based on current experience and knowledge.

Intelligent Agents: Properties (cont)

• Omniscience, Learning, and Autonomy:
  • Omniscient agents know outcomes and can act accordingly.
  • Learning agents gather information and learn from perceived experiences.
  • Autonomous agents adapt to compensate for incorrect prior knowledge.

Intelligent Agents: Nature of Environments

• PEAS (Performance, Environment, Actuators, Sensors):
  • PEAS describes task environments as "problems" with rational agents as "solutions."
• Properties of Task Environments (cont):
  • Fully Observable vs. Partially Observable Environments:
    • fully observable environments are easy to manage, no history is needed;
    • partially observable environments need to track history.
  • Deterministic vs. Stochastic Environments:
    • deterministic environments have outcomes determined by the current state.
  • Episodic vs. Sequential Environments:
    • In sequential environments, prior moves affect future actions (like in checkers).
  • Dynamic vs. Static Environments:
    • Dynamic environments change continuously.
  • Discrete vs. Continuous Environments:
    • discrete environments have finite actions.
  • Known vs. Unknown Environments:
    • in known environments the outcome of each action is known;
    • in unknown environments the agent has to gather knowledge to estimate the actions' outcomes..

Intelligent Agents: Structure of Agents

• Agent = Architecture + Agent Program
  • Architecture: The computing machinery (sensors, actuators) on which the agent operates.
  • Agent Program: The implementation of the agent function.
  • Agent Function: A map from a percept sequence to an action (it defines what to do in a particular situation).

Intelligent Agents: Types of Agents

• Simple Reflex Agents: Agents that respond immediately to percepts using condition-action rules (if-then).
• Model-Based Reflex Agents: Maintain an internal model of the environment over time.
• Goal-Based Agents: Choose actions to achieve goals based on the current state and possible future states.
• Utility-Based Agents: Choose actions based on the expected utility (overall satisfaction).
• Learning Agents: Continuous learning through experience, feedback, performance, and suggestions.

Problem-Solving Agents

• Problem-Solving Agents use sequential procedures to reach a target/goal.
  • Phases of problem-solving:
    • Goal Formulation: Defining goals.
    • Problem Formulation: Refining the goal specification.
    • Search: Finding a solution sequence.
  • A formal problem definition requires initial states, actions, transition models, goal tests, and path costs.
• Examples: 8-puzzle, 8-queens problem.

Example Problems:

• Toy problems as standardized/toy problems, to demonstrate problem-solving techniques - Ex: puzzles.
• Real-world problems that need solutions without requiring detailed descriptions.