Lecture#1-Introduction to AI

8/12/2019 Lecture#1-Introduction to AI

1/74

Department of Computer & Information Sciences

Pakistan Institute of Engineering and Applied Sciences

Intelligence

Chapter 1

1

Umar Faiz

http://www.pieas.edu.pk/umarfaiz

Artificial Intelligence


2/74

Outline

Understand the definition of artificial intelligence

n ers an e eren acu es nvo ve w

intelligent behavior Examine the different ways of approaching AI

Trace briefly the history of AI

Study types of problems that can be currently solved

ability.

Summary

2


3/74

What is AI?

Intelligence A property of mind that encompasses many related

abilities: The capacities to reason, to plan, to solve problems, to think

abstractly, to comprehend ideas, to use language, and tolearn.

Creativity, personality, character, knowledge, or wisdom.

3Source:Wikipedia


4/74

What is AI ?

Artificial Intelligence is concerned with the designo n e gence n an ar c a ev ce.

The term was coined by McCarthy in 1956.

There are two ideas in the definition.1. Intelli ence

2. Artificial device

4


5/74

What is AI ?

What is intelligence? Something that characterizes humans from all other

beings?

Criteria to measure intelli ence or an absolutestandard of judgment for intelligence?

5


6/74

What is AI?

What is intelligence? Regarding intelligence, there are two possibilities:

A system with intelligence is expected to behave asintelligently as a human.

A system with intelligence is expected to behave in the bestpossible manner.

Regarding behavior, are we are interested in The thought process or reasoning ability of the system, or

The final manifestations of the system in terms of its actions.

6


7/74


8/74

Intro to AI

Major Areas of AI Deduction, reasoning, problem solving

Knowledge representation

Learning

Natural language processing Motion and manipulation

Social intelligence

8


9/74

Intro to AI

Tools of AI Search

Logic

Clustering and classification

Neural networks Genetic algorithms

Reasoning tools

9


10/74

Intro to AI

AI Languages Scheme / LISP

Functional

Simple knowledge representation (list)

Prolog

Logic-based

Built-in search engine

Specialized languages Rule lan ua es e. . CLIPS

Planning languages (e.g. STRIPS)

10


11/74

Intro to AI

Definitions four ma or

combinations Based on thinking or

acting.

Based on activity likeSystems

that think

Systems

that think

rational way.

Systems Systems

humans rationally

11


12/74

Intro to AI

1. Acting Humanly Turing Test

Who is Turing? Inventor of modern computers

Turing Thesis Algorithms Turing machines Systems that

think like

Systems that

think

Systems that Systems that

act likehumans

act rationally

12


13/74

Intro to AI

1. Acting Humanly e ur ng es as ree

participants -- two subjects and ajudge. One of the subjects is a

erson and the other is a com uter.Both subjects are hidden from theview of the judge. They communicate

with the judge via text-only channels.

which text channel corresponds tothe human and which corresponds tothe com uter. If the ud e cannot

determine this, then the computerpasses the test.

13


14/74

Intro to AI

An Application of the Turing Test - CAPTCHA:CAPTCHA:

CompletelyAutomatic Public Turing tests to tell Computersand HumansApart

e.g.: Display visually distorted words

Ask user to recognize these words xamp e o app ca on: ave on y umans open ema

accounts

14


15/74

Intro to AI

An Application of the Turing Test - CAPTCHA:

15


16/74

Intro to AI

1. Acting Humanly No program has yet passed Turing test!

(Annual Loebner competition & prize.)

A ro ram that succeeded would need to be ca ableof:

Natural language processing: To enable it to communicatesuccessfully in English.

Knowledge representation: To store what it knows or hears

Automated reasoning: To use the stored information toanswer questions and to draw new conclusions

Machine learning: To adapt to new circumstances and todetect and extrapolate patterns.

16


17/74

Intro to AI

2. Thinking Humanly Try to understand how the mind

works - how do we think? Two ossible routes to find

answers: By introspection - we figure it out

ourselves! By experiment - draw upon

techniques of psychology to conductcontrolled experiments. (Rat in a

Systemsthat thinklike humans

Systems thatthinkrationally

. The discipline of cognitive

science: particularly influential in

Systemsthat act likehumans

Systems thatact rationally

17

v s on, na ura anguage

processing, and learning.


18/74

Intro to AI

2. Thinking Humanly

Human vs Machine Thinking Expert systems - AI success story in early 80's.

'computer program

Rule-based representation of knowledge

Medicine (INTERNIST, MYCIN, . . . )

Geology (PROSPECTOR)

Configuration of computers (R1)

Thinking humanly works!

18


19/74

Intro to AI

2. Thinking Humanly

Human vs Machine Thinking Computer program playing chess

Tried by World champion M. Botvinnik (who also was a

programmer)

Computer way Sophisticated search algorithms

Vast databases Immense computing power

Human world champion beaten!!!

19


20/74

Intro to AI

3. Thinking Rationally Laws of thought approach to AI

Trying to understand how we actually think is one route to AI -but how about how we should think.

Use logic to capture the laws of rational thought as symbols.

Reasoning involves shifting symbols according to well-defined

rules (like algebra). esu s ea se reason ng.

Systemsthat thinklike humans

Systemsthat thinkrationall

Systemsthat act like

humans

Systemsthat act

rationall

20


21/74

Intro to AI

3. Thinking Rationally Logicist approach theoretically attractive.

Lots of problems: Transduction: How to ma the environment to s mbolic

representation;

Representation: How to represent real world phenomena

(time, space, . . . ) symbolically; Reasoning: How to do symbolic manipulation tractably - so it

can be done by real computers!

21


22/74

Intro to AI

4. Acting Rationally Acting rationally = acting to achieve one's goals, given

one's beliefs. Desi n a rational a ent a roach to AI

An agent is just something that acts. Computer agents areexpected to have other attributes that distinguish them from

mere "programs, for example Operating under autonomous control

Perceiving their environment

Persisting over a prolonged time period

Systems that

think likehumans

Systems that

thinkrationall

Systems thatact likehumans

Systems thatact rationally

22


23/74

Intro to AI

4. Acting Rationally Emphasis shifts from designing theoretically best

decision making procedure to best decision makingprocedure possible in circumstances.

Achieving perfect rationality (making the best decisiontheoretically possible) is not usually possible, due to

Limited resources

Limited time

Limited computational power

Limited or uncertain information about environment

The trick is to do the best with what you've got!

23


24/74

Intro to AI

1950 ur ng pre c e a n a ou y years an average

interrogator will not have more than a 70 percent chanceof making the right identification after five minutes ofquestioning".

1957 Newell and Simon predicted that "Within ten years a

computer will be the world's chess champion, unless the" .

24


25/74

Intro to AI

4. Acting Rationally Design a rational agent approach to AI

Rational agent is one that acts so as to achieve the bestoutcome or, when there is uncertainty, the best expectedou come.

Making correct inferences is sometimes part of being a rationalagent, because one way to act rationally is to reason logically tothe conclusion that a given action will achieve one's goals andthen to act on that conclusion.

On the other hand, correct inference is not all of rationality,because there are often situations here there is no provably

, .

25


26/74

Can Machines Act/Think Intelligently?

Yes, if intelligence is narrowly defined as information.

AI has made impressive achievements showing thattasks initially assumed to require intelligence can beau oma e .

But each success of AI seems to push further the limits

of what we consider intelligence.

26


27/74

Typical AI Problems

While studying the typical range of tasks that wem g expec an n e gen en y o per orm, we

need to consider both common tasks as well asex ert tasks.

27


28/74

Typical AI Problems

Common tasks include ecogn z ng peop e, o ec s.

Communicating (through natural language). Navigating around obstacles on the streets.

These tasks are done matter of factly and routinely bypeople and some other animals.

28


29/74

Typical AI Problems

Expert tasks include: Medical diagnosis

Mathematical problem solving

29


30/74

Typical AI Problems

Computer systems have been able to performsop s ca e as s e me ca agnos s,

performing symbolic integration, proving theoremsand la in chess.

However, on the other hand, it has proved to be

very hard to make computer systems performmany rou ne as s a a umans an a o oanimals can do.

without running into things, catching prey and avoidingpredators. Humans and animals are also capable of

30

.


31/74

Intelligent Behaviour

Some of the tasks and applications that shown e gen e av our are:

Perception involving image recognition and computervision

Reasoning

Learning n ers an ng anguage nvo v ng na ura anguageprocessing, speech processing

Solving problems

Robotics

31


32/74

Approaches to AI

Strong AI It aims to build machines that can truly reason and

solve problems. These machines should be self-awareand their overall intellectual ability needs to beindistinguishable from that of a human being.

Strong AI maintains that suitably programmed

machines are ca able of co nitive mental states.

32


33/74

Approaches to AI

Weak AI It deals with the creation of some form of computer-

based artificial intelligence that cannot truly reason andsolve problems, but can act as if it were intelligent.

Weak AI holds that suitably programmed machines cansimulate human cognition.

33


34/74

Approaches to AI

Applied AI It aims to produce commercially viable "smart" systems

For example, a security system that is able to recognise thefaces of people who are permitted to enter a particularu ng.

Applied AI has already enjoyed considerable success.

34


35/74

Approaches to AI

Cognitive AI Computers are used to test theories about how the

human mind works. For example, theories about how we recognise faces and

other objects, or about how we solve abstract problems.

35


36/74

Main Areas of AI

Knowledge representation

earc , espec a y eur s c

search (puzzles, games) Planning

Robotics

Perception

Reasoning under uncertainty,including probabilistic

reasonin

Search

eason ng

Learning

Learning

Agent architectures

Knowledgerep.Planning

Constraintsatisfaction

Robotics and perception Natural language processing

Natural... Expert

36

Systems


37/74

What can AI systems do (limited success) ?

In Com uter vision the s stems are ca able of face reco nition In Robotics, we have been able to make autonomous vehicles.

In Natural language processing, we have systems that are capableof simple machine translation.

Expert systems can carry out medical diagnosis in a narrow domain Speech understanding systems are capable of recognizing several

thousand words continuous speech ann ng an sc e u ng sys ems a een emp oye n sc e u ngexperiments with the Hubble Telescope.

The Learning systems are capable of doing text categorization intoabout a 1000 to ics

In Games, AI systems can play at the Grand Master level in chess(world champion), checkers, etc.

37


38/74

What can AI systems NOT do yet?

n ers an na ura anguage ro us y e.g., rea an

understand articles in a newspaper) Surf the web

Interpret an arbitrary visual scene

Learn a natural language

ons ruc p ans n ynam c rea - me oma ns

Exhibit true autonomy and intelligence

38


39/74

Foundations of AI

The following disciplines contributed ideas,v ewpo n s, an ec n ques o .

Philosophy (428 B .C.-present) Can formal rules be used to draw valid conclusions?

How does the mental mind arise from a physical brain?

Where does knowledge come from?

How does knowled e lead to action?

39


40/74

Foundations of AI

Philosophy (428 B.C .-present) - . .

set of laws governing the rational part of the mind. He developed an informal system for proper reasoning that

allowed one to generate conclusions mechanically, given initialpremises.

40


41/74

Foundations of AI

Philosophy (428 B .C .-present) -

the distinction between mind and matter and of the problemsthat arise.

It held that there is a part of the human mind (or soul or spirit)

that is outside of nature (exempt from physical laws). Animals, on

the other hand, did not possess this dual quality; they could be.

An alternative to dualism is materialism, which holds that thebrain's operation according to the laws of physics constitutes themind.

41


42/74

Foundations of AI

Philosophy (428 B .C .-present) , ,

philosophical doctrine formulated in Vienna in the 1920s,according to which scientific knowledge is the only kind offactual knowledge and all traditional metaphysical doctrinesare to be rejected as meaningless.

42Source: http://www.britannica.com/EBchecked/topic/346336/logical-positivism


43/74


44/74

Foundations of AI

Mathematics (c. 800-present)- , ,

introduced Arabic numerals and algebra.

George Boole (1815-1864) began mathematical development

logic.

Gottlob Frege (1848-1925) extended Boole's logic to include

objects and relations, creating the first-order logic (used todayin basic knowledge representation).

Euclid proposed the first nontrivial algorithm for computinggreatest common denominators.

44

F d ti f AI


45/74

Foundations of AI

Mathematics (c. 800-present) , -

problems that he correctly predicted would occupymathematicians for the bulk of the century.

-theorem showed that in any language expressive enough todescribe theproperties of the natural numbers, there are true

statements that are undecidable in the sense that their truthcanno e es a s e y any a gor m.

45

F d ti f AI


46/74

Foundations of AI

Mathematics (c. 800-present) -

functions are capable of being computed. The Church-Turingthesis, which states that the Turing machine is capable ofcomputing any computable function. Turing also showed thatthere were some functions that no Turing machine cancompute.

46

F d ti f AI


47/74

Foundations of AI

Economics (1776-present)

How should we do this when others may not go along? How should we do this when the payoff may be far in the

47

F d ti f AI


48/74

Foundations of AI

Economics (1776-present)-

treatment of "preferred outcomes7' or utility and was improvedby Frank Ramsey (193 1) and later by John von Neumann andOskar Morgenstern.

Decision theory, which combines probability theory with utilitytheory, provides a formal and complete framework for

decisions (economic or otherwise) made under uncertainty

48

Foundations of AI


49/74

Foundations of AI

Neuroscience (1861-present)

49

Foundations of AI


50/74

Foundations of AI

Neuroscience (1861-present)

Neuroscience is the study of the nervous system, particularlythe brain. The exact wa in which the brain enables thou ht isone of the great mysteries of science.

50

Foundations of AI


51/74

Foundations of AI

Neuroscience (1861-present)' -

brain-damaged patients in 1861 persuaded the medicalestablishment of the existence of localized areas of the brainresponsible for specific cognitive functions.

Hans Berger (1929) invented electroencephalograph (EEG)for the measurement of intact brain activity.

51

Foundations of AI


52/74

Foundations of AI

Psychology (1879-present)

52

Foundations of AI


53/74

Foundations of AI

Psychology (1879-present) -

Wilhelm Wundt (1832-1920) applied the scientific method tothe study of human vision.

Wundt o ened the first laborator of ex erimental s cholo atthe University of Leipzig.

John Watson (1878-1958) initiated Behaviorism movement

that studies objective measures of the percepts (or stimulus)g ven o any an ma an s resu ng ac ons or response . Mental constructs such as knowledge, beliefs, goals, and

reasoning steps were dismissed as unscientific "folk psychology."

53

Foundations of AI


54/74

Foundations of AI

Psychology (1879-present) -

brain as an information-processing device.

Kenneth Craik (1943) specified the three key steps of a-

(1) the stimulus must be translated into an internalrepresentation,

(2) the representation is manipulated by cognitive processeso er ve new n erna represen a ons, an

(3) these are in turn retranslated back into action.

Anderson 1980: A cognitive theory should be like a computer.

54

Foundations of AI


55/74

Foundations of AI

Computer Engineering (1940 present)

ow can we u an e c en compu er

For artificial intelli ence to succeed we need twothings: intelligence and an artifact.

The computer has been the artifact of choice.

55

Foundations of AI


56/74

Foundations of AI


Pascaline: Mechanical adder & substractor (Pascal; mid1600s)

,

Analytic Engine: universal computation; never completed(ideas: addressable memory, stored programs, conditional

jumps) Charles Babbage (1792-1871), Ada Lovelace

56

Foundations of AI


57/74

Foundations of AI


computer built by Alan Turing team in1940, England. Deciphering German messages.

-

Konrad Zuse 1941, Germany

ABC: First electronic computer built by John Atanasoff 1940-

42 US ENIAC: First general-purpose, electronic, digital computer built

by John Mauchy & John Eckert

57

Birth of AI


58/74

Birth of AI

Dartmouth 1956 workshop for 2 months Term artificial intelligence

Fathers of the field introduced

Alan Newell & Herbert Simon

58

Birth of AI


59/74

Birth of AI

Early Enthusiasm (1952-69) Claims: computers can do X

General Problem Solver, Newell & Simon Intentionall solved uzzles in a similar wa as humans do.

Geometry Theorem Prover, Herbert Gelernter, 1959

Arthur Samuels learning checkers program, 1952

, t me s ar ng, v ce ta er: c art y

Integration, IQ geometry problems

, , ,

Adalines [Widrow & Hoff 1960], perceptronconvergence theorem [Rosenblatt 1962]

59

Birth of AI


60/74

A Dose of Reality (1966-74) Simple syntactic manipulation did not scale

Intractability

Perceptrons book with negative result onrepresentation capability of 1-layer ANNs [Minsky &

60

Birth of AI


61/74

Knowledge-based systems (1969-79) DENDRAL: molecule structure identification

[Feigenbaum et al.] Knowledge intensive

Mycin: medical diagnosis [Feigenbaum, Buchanan,Shortliffe]

450 rules knowled e from ex erts no domain theor Better than junior doctors

Certainty factors

Domain knowledge in NLP Knowledge representation: logic, frames...

61

Birth of AI


62/74

AI becomes an industry (1980-88) R1: first successful commercial expert system,

configured computer systems at DEC; saved40M$/year

1988: DEC had 40 expert systems, DuPont 100...

1981: Japans 5th generation project

,Inference, Intellicorp, Teknowledge

LISP-specific hardware: LISP Machines Inc, TI,ym o cs, erox

Industry: few M$ in 1980 -> 2B$ in 1988

62

Birth of AI


63/74

Recent events (1987-) ea -wor app ca ons ra er an oy oma ns

Building on existing work e.g. speech recognition oc, rag e me o s n s

Hidden Markov models now

e.g. planning (unified framework helped progress)

Belief networks & probabilistic reasoning

Reinforcement learning

Multia ent s stems

63

AI Prehistory


64/74

Philosophy Logic, methods of reasoning, mind as physicalsystem foundations of learning, language,rat ona ty

Mathematics Formal representation and proof algorithms,computation, (un)decidability, (in)tractability,robabilit

Economics utility, decision theory Neuroscience physical substrate for mental activity

Psychology phenomena of perception and motor control,exper men a ec n ques Computer building fast computers

engineering

function over time Linguistics knowledge representation, grammar

64

Abridged History of AI


65/74

1943 McCulloch & Pitts: Boolean circuit model of brain 1950 Turin 's "Com utin Machiner and Intelli ence" 1956 Dartmouth meeting: "Artificial Intelligence" adopted

195269 Look, Ma, no hands! 1950s Early AI programs, including Samuel's checkers

program, ewe mon s og c eor st,Gelernter's Geometry Engine

1965 Robinson's complete algorithm for logical reasoning

Neural network research almost disappears

196979 Early development of knowledge-based systems 1980-- AI becomes an industry 1986-- Neural networks return to popularity 1987-- AI becomes a science 1995-- The emergence of intelligent agents

65

State of the art


66/74

IBM Deep Blue:

champion Garry Kasparov in 1997.

66

State of the art


67/74

Proof of Robbins Conjecture:

unsolved for decades.

Dr. Wil liam McCune at

,

his office with computer.The "Proof of Robbins

Conjecture" problem is

67

on the screen.

State of the art


68/74

Autonomous Control:

to keep it following a lane.

It was placed in CMU's NAVLAB computer-controlled minivan-

miles it was in control of steering the vehicle 98% of the time.A human took over the other 2%, mostly at exit ramps.

68

State of the art


69/74

Logistics Planning: ,

planning and scheduling program that involved up to 50,000

vehicles, cargo, and people. NASA's on-board autonomous planning program controlled.

69

State of the art


70/74

Language Understanding and Problem Solving:

puzzles better than most humans, using constraints on

possible word fillers, a large database of past puzzles, and avariety of information sources including dictionaries and onlinedatabases such as a list of movies and the actors that appear.

70

AI and Ethics


71/74

Ethical Concerns: Robot behavior

How can we ensure they do so?

Asimovs Three Laws of Robotics:. , ,

allow a human being to come to harm.

2. A robot must obey orders given it by human beings except

where such orders would conflict with the First Law.3. A robot must protect its own existence as long as suchprotection does not conflict with the First or Second Law.

71

AI and Ethics


72/74

Ethical Concerns: Human behavior

constraints?

As a secondary question, would it be possible to do so?

them from having free will??

Will intelligent systems have consciousness? (Strong AI)

If the do, will it drive them insane to be constrained b artificialethics placed on them by humans?

If intelligent systems develop their own ethics and morality, willwe like what they come up with?

72

Summary


73/74

Different people think of AI differently. Two important

or behavior? Do you want to model humans or work

from an ideal standard? n s oo , we a op e v ew a n e gence s

concerned mainly with rational action. Ideally, anintelligent agent takes the best possible action in as ua on.

73

Summary


74/74

The history of AI has had cycles of success, misplaced,

funding. There have also been cycles of introducing

new creative approaches and systematically refining.

AI has advanced more rapidly in the past decadebecause of greater use of the scientific method inexper men ng w an compar ng approac es.

The subfields of AI have become more integrated, andAI has found common ground with other disciplines.

74

Lecture#1-Introduction to AI

Documents