1.2 - introduction - short
TRANSCRIPT
-
8/8/2019 1.2 - Introduction - Short
1/35
Introduction to AI
-
8/8/2019 1.2 - Introduction - Short
2/35
OutlineWhat is AI ?A brief history of AIState of the art
-
8/8/2019 1.2 - Introduction - Short
3/35
What is AI?AI is a branch of CS with connections topsychology, linguistics, economics, Goal
make artificial systems solve (or help solve) problems thathumans can solve [in the same way as we do??]
Two fundamental aspects:Knowledge representation: making it explicit
Problem solving: heuristic
-
8/8/2019 1.2 - Introduction - Short
4/35
-
8/8/2019 1.2 - Introduction - Short
5/35
What is AI?
Systems that actrationally
Systems that actlike humans
Systems thatthink rationally
Systems that thinklike humans
-
8/8/2019 1.2 - Introduction - Short
6/35
What is AI? - Thinking like humans
Systems that actrationally
Systems that actlike humans
Systems thatthink rationally
Systems that thinklike humans
-
8/8/2019 1.2 - Introduction - Short
7/35
Thinking humanly1960s cognitive revolution
information-processing psychology replaced prevailingorthodoxy of behaviorism
Requires scientific theories of internal activitiesof the brainWhat level of abstraction? knowledge'' (cognitive science)or circuits'' (cognitive neuroscience) ?How to validate? Requires ( top-down ) predicting andtesting behavior of human subjects, or ( bottom-up ) directidentification from neurological data
-
8/8/2019 1.2 - Introduction - Short
8/35
Thinking humanly: Cognitive ScienceThe interdisciplinary field of cognitive science brings together computer models from AI andexperimental techniques from psychology to try
to construct precise and testable theories of theworking of the human mind
-
8/8/2019 1.2 - Introduction - Short
9/35
Thinking humanly: Cogn. NeurosciencMassively parallel neuron networks
Rosenblatts PerceptronMore complex ANN architectures
EvolutionGenetic algorithms (Holland)Evolutionary computation
Artificial lifealife.org
-
8/8/2019 1.2 - Introduction - Short
10/35
What is AI? - Acting like humans
Systems that actrationally
Systems that actlike humans
Systems thatthink rationally
Systems that thinklike humans
-
8/8/2019 1.2 - Introduction - Short
11/35
Acting humanly: The Turing Test
Turing (1950): Computing machinery andintelligence
Can machines think?
Can machines behave intelligently?
-
8/8/2019 1.2 - Introduction - Short
12/35
Acting humanly: The Turing Test
Operational test for intelligent behavior: theImitation Game
-
8/8/2019 1.2 - Introduction - Short
13/35
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 minutesanticipated all major arguments against AI in following 50years
Problemthe Turing test is not reproducible, constructive, or amenable to mathematical analysis
-
8/8/2019 1.2 - Introduction - Short
14/35
Acting humanly: Turing Test
To pass the Turing test, a computer would needto possess the following capabilitiesKnowledge representationAutomated reasoning
Machine learningNatural language (processing and) understanding
In so doing,Turing suggested the maincomponents of AI
-
8/8/2019 1.2 - Introduction - Short
15/35
Acting humanly: Turing test - Eliza
A snapshot of a dialogue between Eliza ( E) andBBNs 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
-
8/8/2019 1.2 - Introduction - Short
16/35
Acting humanly: Turing test - Eliza
A snapshot of a dialogue between Eliza ( E) andBBNs 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 491-1850
Eliza: Weizembaum, 1965
-
8/8/2019 1.2 - Introduction - Short
17/35
Acting humanly: total Turing test
To pass the so-called total Turing test, acomputer would need to possess the followingcapabilities
Knowledge representation
Automated reasoningMachine learningNatural language (processing and) understandingComputer vision
Robotics
-
8/8/2019 1.2 - Introduction - Short
18/35
What is AI? Thinking rationally
Systems that actrationally
Systems that actlike humans
Systems thatthink rationally
Systems that thinklike humans
-
8/8/2019 1.2 - Introduction - Short
19/35
Thinking rationally: Laws of thought
Aristotlewhat are correct arguments / thought processes?his famous syllogisms provided patterns for argumentstructures that always give correct conclusions givencorrect premises
The development of formal logic ...in the late nineteenth and early twentieth centuriesprovided a precise notation for reasoning about facts
Two main obstaclesnot easy to take informal knowledge and state it in theformal terms required by logical notationbig difference between being able to solve a problem inprinciple and doing so in practice
-
8/8/2019 1.2 - Introduction - Short
20/35
What is AI? - Acting rationally
Systems that actrationally
Systems that actlike humans
Systems thatthink rationally
Systems that thinklike humans
-
8/8/2019 1.2 - Introduction - Short
21/35
Acting rationally: Rational behaviorRational behavior: doing the right thing
The right thing: that which is expected to maximize goalachievement, 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 andpursuit, is thought to aim at some good
-
8/8/2019 1.2 - Introduction - Short
22/35
Acting rationally: Rational agentsWe want to design rational agentsAn agent is an entity that perceives and actsAbstractly, an agent is a function from percepthistories to actions: f: P A
-
8/8/2019 1.2 - Introduction - Short
23/35
Acting rationally: Rational agentsFor any given class of environments and tasks,we seek the agent (or class of agents) with thebest performance
Caveat: computational limitations make perfectrationality unachievable design best programfor given machine resourcesLimited or bounded rationality: optimize vssatisfy (Herbert Simon)
-
8/8/2019 1.2 - Introduction - Short
24/35
Acting rationally: EngineeringRobotics (building artifacts that work)
Control: differential equations vs learningActing: mechanics of effectors (arms, legs, )Perception: sensors, vision, speech recognition...
-
8/8/2019 1.2 - Introduction - Short
25/35
An example: Mars Exploration RoversSeveral generations of rovers: Pathfinder andSojourner 1997, Spirit and Opportunity 2003
mars.jpl.nasa.gov/mer
-
8/8/2019 1.2 - Introduction - Short
26/35
Another example: RobocupSimulation and real robotsMulti-agent cooperationAn annual event
www.robocup.org
-
8/8/2019 1.2 - Introduction - Short
27/35
AI and CS
Automatic problem solvingExpert systemsHeuristic programmingUncertainty management and reasoning
New programming paradigmsObject orientedFunctionalLogicWeb agents
Natural language analysis and synthesis
-
8/8/2019 1.2 - Introduction - Short
28/35
AI: PrehistoryPhilosophy logic, methods of reasoning
mind as physical systemfoundations of learning, language, rationality
Mathematics formal representation and proof
algorithms, computation, (un)decidabilityprobabilityPsychology adaptation
phenomena of perception and motorial control
experimental techniques (psychophysics, etc.)
-
8/8/2019 1.2 - Introduction - Short
29/35
AI: PrehistoryEconomics formal theory of rational decisionsGames games theoryLinguistics knowledge representation
grammar
Neuroscience physical substrate for mental activityControl theory homeostatic systemsstabilitysimple optimal agent designs
-
8/8/2019 1.2 - Introduction - Short
30/35
AI: Some points in time1943 McCulloch & Pitts: Boolean circuit model of
brain1950 Turing's Computing Machinery and
Intelligence article
1952-69 Look, Ma, no hands! (early enthusiasm)1950s Early AI programs:- Samuel's checkers program,- Newell & Simon's Logic Theorist,
- Gelernter's Geometry Engine1956 Dartmouth meeting: term Artificial
Intelligence adopted
-
8/8/2019 1.2 - Introduction - Short
31/35
AI: Some points in time1960s Lisp (McCarthy)1965 Robinson's complete algorithm for logical
reasoning1966-74 AI discovers computational complexity
Artificial Neural Networks research almostdisappears1969-79 Early development of knowledge-based systems
(Buchanan & Shortliffe)
Prolog (Colmerauer)1980-88 Expert systems industry booms
Machine learning1988-93 Expert systems industry busts: AI Winter
-
8/8/2019 1.2 - Introduction - Short
32/35
AI: Some points in time1985-95 Artificial Neural Networks return to popularity1988-now Resurgence of probability; technical depth
Nouvelle AI'': ALife, GAs, soft computing1995-now Agents, ... agents everywhere
Data mining, SoftbotsChess: Deep Thought vs Kasparov
-
8/8/2019 1.2 - Introduction - Short
33/35
AI: State of the art
Autonomous planning and schedulingGame playingAutonomous controlMedical diagnosis based on probabilistic analysisLogistics planningRobotic microsurgeryLanguage understanding and problem solving...
-
8/8/2019 1.2 - Introduction - Short
34/35
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)
h f
-
8/8/2019 1.2 - Introduction - Short
35/35
AI: The future you name it
Intelligent homesPersonalized assistants in cell phones and PDAsRoboticsDynamically created moviesSoccer between humans and robots...