Introduction to Artificial Intelligence
Artificial Intelligence
Jesus A. Gonzalez
August 25, 2016
- These slides and most of the figures are based or taken from the book:
- Artificial Intelligence: A Modern Approach
- Stuart Russell and Peter Norvig
- Third Edition, Prentice Hall
Introduction to Artificial Intelligence
Introduction to AI
- Importance of Man’s intelligence
- Man’s Intelligence
- We have tried to understand how we think
- percieve, understand, predict, manipulate the world
- Artificial Intelligence
- Focused not only to understand intelligence but to Build intelligent entities
- Examples of AI in many subfields
- General: learning, perception
- Specific: playing chess, theorem’s proving, autonomous driving, diagnosing diseases
What is AI?
- 4 approaches
- Top: Concerned with thought processes and reasoning
- Bottom: Concerned with behavior
- Left: Success in terms of fidelity to human performance
- Right: Success in terms of ideal performance, rationality
What is AI?
- Thinking Humanly
- “The exciting new effort to make computers think … machines with minds, in the full and literal sense.” (Haugeland, 1985)
- “[The automation of] activities that we associate with human thinking, activities such as decision-making, problem solving, learning …” (Vellman, 1978)
What is AI?
- Acting Humanly
- “The art of creating machines that perform functions that require intelligence when performed by people.” (Kurzweil, 1990)
- “The study of how to make computers do things at which, at the moment, people are better.” (Rich and Knight, 1991)
What is AI?
- Thinking Rationally
- “The study of mental faculties through the use of computational models.” (Charniak and McDermott, 1985)
- “The study of the computations that make it possible to perceive, reason, and act.” (Winston, 1992)
What is AI?
- Acting Rationally
- “Computational Intelligence is the study of the design of intelligent agents.” (Poole et al., 1998)
- “AI … is concerned with intelligent behavior in artifacts.” (Nilsson, 1998)
Acting Humanly
- The Turing Test Approach
- Turing Test, Alan Turing (1950), “Computer Machinery and Intelligence”
- Designed to define intelligence
- Computer passes test if…
- Human interrogator, cannot distinguish between person and computer after asking a set of questions
Acting Humanly
Acting Humanly
- Turing Test
- A lot of work involved in such a computer program
- NLP (communicate in English)
- Knowledge representation (store knowledge)
- Automated reasoning (use information to answer questions, create new conclusions)
- Machine learning (adapt to new circusnstances, find patterns)
Acting Humanly
- The Total Turing Test
- Includes video signal
- Interrogator test the subject’s perceptual abilities
- Interrogator may pass physical objects “through the hatch”
- Passing the Total Turing Test
- Computer vision (perceive objects)
- Robotics (manipulate objects and move them)
Thinking Humanly
- The Cognitive Modeling Approach
- Say that a given program thinks like a human
- How do humans think?
- How the human mind works?
- Three ways to do it
- Introspection - Catch our thoughts
- Psychological experiments - Observe person in action
- Brain imaging - Observe the brain in action
- Express theory of the mind as a computer program
Thinking Humanly
- Does the program’s input-output behavior matches the human’s input-output behavior?
- Program’s mechanisms could be operating in humans
- Herbert Simon developed “General Problem Solver” (GPS), Newell and Simon, 1961
- Compare trace of GPS reasoning steps to traces of human subjects solving same problem
Thinking Humanly
Thinking Humanly
- Cognitive Science: AI Computer Models + Psychology Experimental Techniques to construct precise and testable theories of the human mind
- Based on experimental investigation of actual humans or animals
Thinking Rationally
- The “laws of thought” Approach
- Aristotle, tried to codify “right thinking”, an irrefutable reasoning process
- Syllogisms –> patterns for argument structures
- Obtain correct conclussions when given correct premises
- Socrates is a man; all men are mortal; therefore, Socrates is mortal
- Laws of thought suppossed to govern the operation of mind
- Initiated the field of Logic
Thinking Rationally
- 19th century, logicians developed notation for statement of all kinds of objects in the world and relations among them
- The field of Logic
- 1965, programs could (in principle) solve any solvable problem described in logical notation
- The logicist tradition within AI, use logic-programs to create intelligent systems
- Obstacles in this approach
- Not easy to take informal knowledge and state it formally, even more with uncertainty
- Different solving a problen in principle and in practice
- Computational complexity (i.e. large search spaces)
- Computer resources (i.e. memory)
Acting Rationally
- The Rational Agent Approach
- An agent is something that acts
- From Latin agere, to do
- Operates autonomously
- Perceives their environment
- Persist over a prolonged time period
- Adapt to change
- Create and pursue goals
Acting Rationally
- Rational Agent
- Acts to achieve the best outcome
- Under uncertainty: acts to achieve the best expected outcome
- Requires more than the laws of thought (more than only correct inferences)
- Ways of acting rationally not involving inference
- Recoiling from a hot stove is a reflex
- Better if action taken quickly than with careful deliberation
Acting Rationally
- Advantages
- More general than the laws of thught approach
- Inference is one of several ways to achieve rationality
- More amenable to scientific development than approaches based on human behavior or human thought
- Rationality defned mathematically is general
- Generate different designs to achieve rationality
Foundations of Artificial Intelligence
AI prehistory
- Philosophy
- Logic
- Methods of reasoning
- Mind as physical system
- Foundations of learning
- Language
- Rationality
AI prehistory
- Mathematics
- Formal representation and proof
- Algorithms
- Computation
- (un)decidability
- (in)tractability
- Probability
AI prehistory
- Psychology
- Adaptation
- Behaviorism
- Phenomena of perception and motor control
- Experimental techniques (psychophysics, etc)
AI prehistory
- Economics
- Formal theory of rational decisions
- Utility
- Decision theory
- Game theory
- Operations Research
- Satisficing
AI prehistory
- Linguistics
- Knowledge representation
- Grammar
AI prehistory
- Neuroscience
- Plastic physical substrate for mental activity
- Neuron
AI prehistory
- Control Theory
- Homeostatic systems
- Stability
- Simple optimal agent designs
- Cybernetics
- Objective function
The History of Artificial Intelligence
History of AI
- Gestation of AI (1943 - 1955)
- 1943 McCulloch & Pitts: Boolean circuit model of brain
- 1950 Turing’s “Computing Machinery and Intelligence”
- 1952-69 Look, Ma, no hands!
- 1950’s Early AI programs
- Samuel’s checkers program
- Newell & Simon’s Logic Theorist
- Gelernter’s Geometry Engine
History of AI
- Birth of AI (1956)
- John McCarthy, Minsky, Claude Shannon, Nathaniel Rochester, and more
- Darmouth College meeting
- Official birthplace of the field
- “Artificial Intelligence” adopted
History of AI
- 1965 Robinson’s complete algorithm for logical reasoning
- 1966-74 AI discovers computational complexity, Neural network research almost disappears
- 1969-79 Early development of knowledge-based systems
- 1980-93 Expert systems industry busts: “AI Winter”
- 1985-95 Neural networks return to popularity
- 1988- Resurgence of probability, HMM’s, Data Mining, Bayesian Networks, general increase in technical depth, “Nouvelle AI”: ALife, GAs, soft computing
- 1995- Agents everywhere…
- 2003- Human-level AI back on the agenda
- 2001- Availability of very large data sets
State of the Art in Artificial Intelligence
State of the Art in AI
State of the Art in AI
State of the Art in AI
- Autonomous planning and scheduling
State of the Art in AI
- Game playing
- IBM’s Deep Blue wins in Chess (vs Gary Kasparov)
- IBM’s Watson wins in Jeopardy
State of the Art in AI
State of the Art in AI
- Logistics planning
- Dynamic Analysis and Replanning Tool, DART (Cross and Walker, 1994)
- Automated logistics planning and scheduling for transportation
- During the Persian Gulf crisis of 1991
- 50,00 vehicles, cargo and people at a time
Assignment
- Exercises: 1.1, 1.7, 1.14 from the textbook
- Artificial Intelligence: A Modern Approach
- Stuart Russell and Peter Norvig
- Third Edition, Prentice Hall
- Due date: June 13, 2016 at the end of the class
