ARTIFICIAL ARTIFICIAL INTELLIGENCEINTELLIGENCE

Introduction: Chapter 1Introduction: Chapter 1

20272027

Textbook: S. Russell and P. Norvig Textbook: S. Russell and P. Norvig Artificial Artificial Intelligence: A Modern Approach Intelligence: A Modern Approach Prentice Hall, Prentice Hall, 2003, 2003, Second EditionSecond Edition

Lecturer: Shih-Hung WuLecturer: Shih-Hung Wu Grading: Homework (20%), Quiz (15%), Grading: Homework (20%), Quiz (15%), Midterm test (20%), Final exam (25%), Project Midterm test (20%), Final exam (25%), Project

(20%) (20%) Note that attendance at every lecture will be Note that attendance at every lecture will be

taken and constitutes part of the class taken and constitutes part of the class participation grade.participation grade.

Midterm test (in class, 2 hr) are open-bookMidterm test (in class, 2 hr) are open-book

OutlineOutline

Course overviewCourse overview What is AI?What is AI? A brief historyA brief history The state of the artThe state of the art

Course overviewCourse overview

Introduction and Agents (chapters 1,2)Introduction and Agents (chapters 1,2) Search (chapters 3,4,5,6)Search (chapters 3,4,5,6) Logic (chapters 7,8,9)Logic (chapters 7,8,9) Planning (chapters 11,12)Planning (chapters 11,12) Uncertainty (chapters 13,14)Uncertainty (chapters 13,14) Learning (chapters 18,20)Learning (chapters 18,20) Natural Language Processing (chapter Natural Language Processing (chapter

22,23)22,23)

What is AI?What is AI?

Views of AI fall into four categories:Views of AI fall into four categories:

Thinking humanlyThinking humanly Thinking Thinking rationally rationally

Acting humanlyActing humanly Acting rationally Acting rationally

The textbook advocates "acting The textbook advocates "acting rationally"rationally"

Acting humanly: Turing Acting humanly: Turing TestTest

Turing (1950) "Computing machinery and intelligence":Turing (1950) "Computing machinery and intelligence": "Can machines think?" "Can machines think?" "Can machines behave "Can machines behave

intelligently?"intelligently?" Operational test for intelligent behavior: the Imitation Operational test for intelligent behavior: the Imitation

GameGame

Predicted that by 2000, a machine might have a 30% Predicted that by 2000, a machine might have a 30% chance of fooling a lay person for 5 minuteschance of fooling a lay person for 5 minutes

Anticipated all major arguments against AI in following 50 Anticipated all major arguments against AI in following 50 yearsyears

Suggested major components of AI: knowledge, reasoning, Suggested major components of AI: knowledge, reasoning, language understanding, learninglanguage understanding, learning

Thinking humanly: Thinking humanly: cognitive modelingcognitive modeling

1960s "cognitive revolution": information-1960s "cognitive revolution": information-processing psychology processing psychology

Requires scientific theories of internal Requires scientific theories of internal activities of the brainactivities of the brain

-- How to validate? Requires -- How to validate? Requires 1) Predicting and testing behavior of human 1) Predicting and testing behavior of human

subjects (top-down)subjects (top-down) or 2) Direct identification from neurological or 2) Direct identification from neurological

data (bottom-up)data (bottom-up) Both approaches (roughly, Cognitive Both approaches (roughly, Cognitive

Science and Cognitive Neuroscience) Science and Cognitive Neuroscience) are now distinct from AIare now distinct from AI

Thinking rationally: Thinking rationally: "laws of thought""laws of thought"

Aristotle: what are correct arguments/thought Aristotle: what are correct arguments/thought processes?processes?

Several Greek schools developed various forms Several Greek schools developed various forms of of logiclogic: : notationnotation and and rules of derivationrules of derivation for for thoughts; may or may not have proceeded to thoughts; may or may not have proceeded to the idea of mechanizationthe idea of mechanization

Direct line through mathematics and Direct line through mathematics and philosophy to modern AIphilosophy to modern AI

Problems: Problems: 1.1. Not all intelligent behavior is mediated by logical Not all intelligent behavior is mediated by logical

deliberationdeliberation2.2. What is the purpose of thinking? What thoughts What is the purpose of thinking? What thoughts

should I have?should I have?

3.3.

Acting rationally: Acting rationally: rational agentrational agent

RationalRational behavior: doing the right behavior: doing the right thingthing

The right thing: that which is expected The right thing: that which is expected to maximize goal achievement, given to maximize goal achievement, given the available informationthe available information

Doesn't necessarily involve thinking – Doesn't necessarily involve thinking – e.g., blinking reflex – but thinking e.g., blinking reflex – but thinking should be in the service of rational should be in the service of rational actionaction

Rational agentsRational agents An An agentagent is an entity that perceives and acts is an entity that perceives and acts This course is about designing rational agentsThis course is about designing rational agents Abstractly, an agent is a function from Abstractly, an agent is a function from

percept histories to actions:percept histories to actions:[[ff: : P*P* AA]]

For any given class of environments and For any given class of environments and tasks, we seek the agent (or class of agents) tasks, we seek the agent (or class of agents) with the best performancewith the best performance

Caveat: computational limitations make Caveat: computational limitations make perfect rationality unachievableperfect rationality unachievable design best design best programprogram for given machine resources for given machine resources

AI prehistoryAI prehistory PhilosophyPhilosophy Logic, methods of reasoning, mind as Logic, methods of reasoning, mind as

physical physical system foundations of learning, language,system foundations of learning, language,

rationalityrationality MathematicsMathematics Formal representation and proof algorithms,Formal representation and proof algorithms,

computation, (un)decidability, (in)tractability,computation, (un)decidability, (in)tractability,probabilityprobability

EconomicsEconomics utility, decision theory utility, decision theory NeuroscienceNeuroscience physical substrate for mental activityphysical substrate for mental activity Psychology Psychology phenomena of perception and motor control,phenomena of perception and motor control,

experimental techniquesexperimental techniques Computer Computer building fast computers building fast computers

engineeringengineering Control theoryControl theory design systems that maximize an objectivedesign systems that maximize an objective

function over time function over time LinguisticsLinguistics knowledge representation, grammarknowledge representation, grammar

Abridged history of AIAbridged history of AI 1943 1943 McCulloch & Pitts: Boolean circuit model of brainMcCulloch & Pitts: Boolean circuit model of brain 1950 1950 Turing's "Computing Machinery and Intelligence"Turing's "Computing Machinery and Intelligence" 19561956 Dartmouth meeting: "Artificial Intelligence" adoptedDartmouth meeting: "Artificial Intelligence" adopted 1952—691952—69 Look, Ma, no hands! Look, Ma, no hands! 1950s1950s Early AI programs, including Samuel's checkersEarly AI programs, including Samuel's checkers

program, Newell & Simon's Logic Theorist, program, Newell & Simon's Logic Theorist, Gelernter's Geometry EngineGelernter's Geometry Engine

19651965 Robinson's complete algorithm for logical reasoningRobinson's complete algorithm for logical reasoning 1966—731966—73 AI discovers computational complexityAI discovers computational complexity

Neural network research almost disappearsNeural network research almost disappears 1969—791969—79 Early development of knowledge-based systemsEarly development of knowledge-based systems 1980-- 1980-- AI becomes an industry AI becomes an industry 1986-- 1986-- Neural networks return to popularityNeural networks return to popularity 1987--1987-- AI becomes a science AI becomes a science 1995--1995-- The emergence of intelligent agents The emergence of intelligent agents

State of the artState of the art Deep Blue defeated the reigning world chess Deep Blue defeated the reigning world chess

champion Garry Kasparov in 1997 champion Garry Kasparov in 1997 Proved a mathematical conjecture (Robbins Proved a mathematical conjecture (Robbins

conjecture) unsolved for decades conjecture) unsolved for decades No hands across America (driving autonomously No hands across America (driving autonomously

98% of the time from Pittsburgh to San Diego) 98% of the time from Pittsburgh to San Diego) During the 1991 Gulf War, US forces deployed an During the 1991 Gulf War, US forces deployed an

AI logistics planning and scheduling program that AI logistics planning and scheduling program that involved up to 50,000 vehicles, cargo, and people involved up to 50,000 vehicles, cargo, and people

NASA's on-board autonomous planning program NASA's on-board autonomous planning program controlled the scheduling of operations for a controlled the scheduling of operations for a spacecraft spacecraft

ProverbProverb solves crossword puzzles better than solves crossword puzzles better than most humansmost humans