Outline. What is AI? A brief history of AI State of the art

Transcription

1 Introduction to AI

2 Outline What is AI? A brief history of AI State of the art

3 What is AI? AI is a branch of CS with connections to psychology, linguistics, economics, Goal make artificial systems solve (or help solve) problems that humans can solve [in the same way as we do??] Two fundamental aspects: Knowledge representation: making it explicit Problem solving: heuristic

4 What is AI? Artificial Intelligence Philosophy Mathematics Linguistics Psychology Art Engineering Biology Comp. Science

5 What is AI? Systems that think like humans Systems that think rationally Systems that act like humans Systems that act rationally

6 What is AI? - Thinking like humans Systems that think like humans Systems that think rationally Systems that act like humans Systems that act rationally

7 Thinking humanly 1960s cognitive revolution information-processing psychology replaced prevailing orthodoxy of behaviorism Requires scientific theories of internal activities of the brain What level of abstraction? knowledge'' (cognitive science) or circuits'' (cognitive neuroscience)? How to validate? Requires (top-down) predicting and testing behavior of human subjects, or (bottom-up) direct identification from neurological data

8 Thinking humanly: Cognitive Science The interdisciplinary field of cognitive science brings together computer models from AI and experimental techniques from psychology to try to construct precise and testable theories of the working of the human mind

9 Thinking humanly: Cogn. Neuroscience Massively parallel neuron networks Rosenblatt s Perceptron More complex ANN architectures Evolution Genetic algorithms (Holland) Evolutionary computation Artificial life alife.org

10 What is AI? - Acting like humans Systems that think like humans Systems that think rationally Systems that act like humans Systems that act rationally

11 Acting humanly: The Turing Test Turing (1950): Computing machinery and intelligence Can machines think? Can machines behave intelligently?

12 Acting humanly: The Turing Test Operational test for intelligent behavior: the Imitation Game

13 Acting humanly: The Turing Test Alan Turing... predicted that by 2000, a machine might have a 30% chance of fooling a lay person for 5 minutes anticipated all major arguments against AI in following 50 years Problem the Turing test is not reproducible, constructive, or amenable to mathematical analysis

14 Acting humanly: Turing Test To pass the Turing test, a computer would need to possess the following capabilities Knowledge representation Automated reasoning Machine learning Natural language (processing and) understanding In so doing,turing suggested the main components of AI

15 Acting humanly: Turing test - Eliza A snapshot of a dialogue between Eliza (E) and BBN s Vice President (V) V: Do you think that I can use the computer this morning? E: Why do you ask? V: I might be able to make some additional Telcom sales. E: Why aren't you so sure? V: My perspective customers have never seen a demonstration of the system. E: Does this mean very much to you? Eliza: Weizembaum, 1965

16 Acting humanly: Turing test - Eliza A snapshot of a dialogue between Eliza (E) and BBN s Vice President (V) [continued] V: Of course it does. E: You seem very positive about that. V: This is ridiculous. E: What do you mean, this is ridiculous? V: Is there any reason why I can't use the computer now? E: Tell me more about your perspective customers. V: Please dial me up on Eliza: Weizembaum, 1965

17 Acting humanly: total Turing test To pass the so-called total Turing test, a computer would need to possess the following capabilities Knowledge representation Automated reasoning Machine learning Natural language (processing and) understanding Computer vision Robotics

18 What is AI? Thinking rationally Systems that think like humans Systems that think rationally Systems that act like humans Systems that act rationally

19 Thinking rationally: Laws of thought Aristotle what are correct arguments / thought processes? his famous syllogisms provided patterns for argument structures that always give correct conclusions given correct premises The development of formal logic... in the late nineteenth and early twentieth centuries provided a precise notation for reasoning about facts Two main obstacles not easy to take informal knowledge and state it in the formal terms required by logical notation big difference between being able to solve a problem in principle and doing so in practice

20 What is AI? - Acting rationally Systems that think like humans Systems that think rationally Systems that act like humans Systems that act rationally

21 Acting rationally: Rational behavior Rational behavior: doing the right thing The right thing: that which is expected to maximize goal achievement, given the available information Does not necessarily involve thinking (e.g., blinking reflex)... but thinking should be in the service of rational action Aristotle (Nicomachean Ethics): Every art and every inquiry, and similarly every action and pursuit, is thought to aim at some good

22 Acting rationally: Rational agents We want to design rational agents An agent is an entity that perceives and acts Abstractly, an agent is a function from percept histories to actions: f: P A

23 Acting rationally: Rational agents For any given class of environments and tasks, we seek the agent (or class of agents) with the best performance Caveat: computational limitations make perfect rationality unachievable design best program for given machine resources Limited or bounded rationality: optimize vs satisfy (Herbert Simon)

24 Acting rationally: Engineering Robotics (building artifacts that work) Control: differential equations vs learning Acting: mechanics of effectors (arms, legs, ) Perception: sensors, vision, speech recognition...

25 An example: Mars Exploration Rovers Several generations of rovers: Pathfinder and Sojourner 1997, Spirit and Opportunity 2003 mars.jpl.nasa.gov/mer

26 Another example: Robocup Simulation and real robots Multi-agent cooperation An annual event

27 AI and CS Automatic problem solving Expert systems Heuristic programming Uncertainty management and reasoning New programming paradigms Object oriented Functional Logic Web agents Natural language analysis and synthesis

28 AI: Prehistory Philosophy Mathematics Psychology logic, methods of reasoning mind as physical system foundations of learning, language, rationality formal representation and proof algorithms, computation, (un)decidability probability adaptation phenomena of perception and motorial control experimental techniques (psychophysics, etc.)

29 AI: Prehistory Economics formal theory of rational decisions Games games theory Linguistics knowledge representation grammar Neuroscience physical substrate for mental activity Control theory homeostatic systems stability simple optimal agent designs

30 AI: Some points in time 1943 McCulloch & Pitts: Boolean circuit model of brain 1950 Turing's Computing Machinery and Intelligence article Look, Ma, no hands! (early enthusiasm) 1950s Early AI programs: - Samuel's checkers program, - Newell & Simon's Logic Theorist, - Gelernter's Geometry Engine 1956 Dartmouth meeting: term Artificial Intelligence adopted

31 AI: Some points in time 1960 s Lisp (McCarthy) 1965 Robinson's complete algorithm for logical reasoning AI discovers computational complexity Artificial Neural Networks research almost disappears Early development of knowledge-based systems (Buchanan & Shortliffe) Prolog (Colmerauer) Expert systems industry booms Machine learning Expert systems industry busts: AI Winter

32 AI: Some points in time Artificial Neural Networks return to popularity 1988-now Resurgence of probability; technical depth Nouvelle AI'': ALife, GAs, soft computing 1995-now Agents,... agents everywhere Data mining, Softbots Chess: Deep Thought vs Kasparov

33 AI: State of the art Autonomous planning and scheduling Game playing Autonomous control Medical diagnosis based on probabilistic analysis Logistics planning Robotic microsurgery Language understanding and problem solving...

34 AI: Schools Problem solving (Simon, Newell) Simple agent societies (Minsky, Brooks) Robotics (Nilsson) Language and representation (Shank) Common sense reasoning (McCarthy, Lenat) Evolutionary Computation (Holland, Koza) Artificial Neural Networks (McCulloch, Pitts) Expert systems (Buchanan, Shortliffe) Machine learning (Samuel, Mitchell) Logic (Robinson, Colmerauer)

35 AI: The future you name it Intelligent homes Personalized assistants in cell phones and PDAs Robotics Dynamically created movies Soccer between humans and robots...