WHAT IS ARTIFICIAL INTELLIGENCE (AI)?

Other Definitions

• The Turing Test (Alan Turing, 1950)

• The science and engineering of making intelligent machines, especially intelligent
computer programs [John McCarthy, Dartmouth conference 1956]
Dimensions of AI (4 Views)
Dimensions of AI (4 Views)

1. Thinking rationally
- formalize the inference process
2. Thinking humanly
- model human cognition
3. Acting humanly
- exhibit human behavior
4. Acting rationally
- doing the right thing
1. Thinking rationally: Laws of Thought

• Aristotle: what are correct arguments/thought processes?

• Several Greek schools developed various forms of logic:
• notation and rules of derivation for thoughts;
• Direct line through mathematics and philosophy to modern AI

Example (Aristotle's Syllogism)

All computers use energy
Using energy always generates heat
Therefore, All computers generate heat
Problems:
• Not all intelligent behavior is mediated by logical deliberation
e.g., Informal knowledge is not precise. And therefore, under a logical premise, it is difficult to
model uncertainty.
2. Thinking humanly: Cognitive Science
• Model human cognition
• Requires scientific theories of internal activities of the brain
• The General Problem Solver developed in 1957 by Alan Newell and Herbert
Simon thought of a single computer program that would solve any problem ;
given a sustainable description of the problem (grand vision !!!).
– How to validate? Requires
1) Predicting and testing behavior of human subjects
or 2) Direct identification from neurological data
- Can't verify formally about the decision that machine takes (Big Issue!!)
• Both approaches (roughly, Cognitive Science and Cognitive Neuroscience) are
now distinct from AI
 3. Acting humanly: The Turing test
• “Can machines behave intelligently?”
- exhibiting human behavior
• Creating machines that perform functions that require intelligence when the same
functions are performed by people
• Operational test for intelligent behavior: the Imitation Game or Turing Test (1950)

Suggested major components of AI: natural language processing, knowledge representation,

automated reasoning, machine learning, computer vision, robotics
Problem: Turing test is not reproducible, constructive, or amenable to mathematical analysis
4. Acting rationally
• Rational behavior: doing the right thing [Informally]
• The term rational in a very specific, technical way:
• Rational: maximally achieving pre-defined goals, given the available information
• Rationality only concerns what decisions are made (not the thought process
behind them)
• Goals are expressed in terms of the utility of outcomes
• Being rational means maximizing your expected utility
• Expected utility is the utility of an action or event over a time period when the
circumstances are unknown
• Computational limitations make perfect rationality unachievable
WHAT AI CAN DO?

1. AUTOMATED PROBLEM SOLVING

• SEARCH
• Enormous computational complexity
• Use of domain knowledge- heuristics
• PARADIGMS
• Dynamic programming
• Heuristic Search
• Evolutionary Algorithms
WHAT AI CAN DO?

2. KNOWLEDGE AND DEDUCTION

• How to store and retrieve knowledge?
• How to interpret facts and rules, and be able to deduce?
• Propositional logic, FOL
• PARADIGMS
• Knowledge-based systems
• Expert systems
• Formal Verification [for Safety-critical, Security-critical, Financial, Legal]
WHAT AI CAN DO?

3. LEARNING
• Can we learn to solve a problem better?
• Learning the answers
• Learning the rules of the game
• Learning to plan
PARADIGMS
• Belief (Bayesian) Networks
• Perceptrons and neural networks
WHAT THEN IS AI?

ARTIFICIAL INTELLIGENCE = AUTOMATED PROBLEM SOLVING

+
LOGIC AND DEDUCTION
+
MACHINE LEARNING

A basic computational background which is necessary to make AI real

THE AI PROBLEMS

Mundane Tasks
• Perception
• Vision
• Speech
• Touch
• Natural Language
• Understanding
• Generation
• Translation
• Planning
• Commonsense Reasoning
• Navigating
AI TASKS

Formal Tasks
• Games
• Chess [Deep Blue, CMU/IBM] [Loss, G. Kasparov, 1996 ][Win, G. Kasparov, 1997]
• Backgammon
• Checkers
• Go [AlphaGo, DeepMind-Google] [Win, Lee Sedol, 2016]
• Mathematics
• Geometry
• Logic
• Integral Calculus
• Proving properties of programs
AI TASKS
• Expert Tasks
• Engineering
• Design
• Fault finding
• Manufacturing planning
• Scientific Analysis
• Medical Diagnosis
• Financial Analysis
WHAT'S EASY AND WHAT'S HARD?
• It has been easier to mechanize many of the high-level
(Formal/Expert) tasks we usually associate with "intelligence" in
people
• It has been hard to mechanize tasks that lots of animals can do
• Walking around without running into things
• Caching prey and avoiding predators
• Interpreting complex sensory information
• Modeling the internal states of other animals from their behavior
A (Short) History of AI
• 1940-1950: Early days
• 1943: McCulloch & Pitts: Boolean circuit model of brain
• 1950: Turing's “Computing Machinery and Intelligence”
• 1950—70: Excitement
• 1950s: Early AI programs, including Samuel's checkers
program, Newell & Simon's Logic Theorist, Gelernter's
Geometry Engine
• 1956: Dartmouth meeting: “Artificial Intelligence” adopted
• 1965: Robinson's complete algorithm for logical reasoning
A (Short) History of AI (cont..)

• 1970—90: Knowledge-based approaches

• 1969—79: Early development of knowledge-based
systems
• 1980—88: Expert systems industry booms
• 1988—93: Expert systems industry busts: “AI Winter”
• 1990—: Statistical approaches
• Resurgence of probability, focus on uncertainty
• General increase in technical depth
• Agents and learning systems... “AI Spring”?
• 1996: Kasparov defeats Deep Blue at chess
• 1997: Deep Blue defeats Kasparov at chess
A (Short) History of AI (cont..)

• 2000—: Where are we now?

▪ Big data, big compute, neural networks
▪ Some re-unification of sub-fields
▪ AI used in many industries
▪ Chess engines running on ordinary laptops can
defeat the world’s best chess players
▪ 2011: IBM’s Watson defeats Ken Jennings and
Brad Rutter at Jeopardy!
▪ 2016: Google’s AlphaGo beats Lee Sedol at Go
A (Short) History of AI (cont..)