Department of Computer & Information Sciences

Pakistan Institute of Engineering and Applied Sciences

Introduction to Artificial Introduction to Artificial
Intelligence
Chapter 1
1
Umar Faiz
http://www.pieas.edu.pk/umarfaiz
Artificial Intelligence
Outline
Understand the definition of artificial intelligence
U d t d th diff t f lti i l d ith Understand the different faculties involved with
intelligent behavior
Examine the different ways of approaching AI
Trace briefly the history of AI
Study types of problems that can be currently solved
by computers and those that are as yet beyond its by computers and those that are as yet beyond its
ability.
Summary
2
What is AI?
Intelligence
A f i d h l d 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 to
learn.
Creativity, personality, character, knowledge, or wisdom.
3
Source:Wikipedia
What is AI ?
Artificial Intelligence is concerned with the design
of intelligence in an artificial de ice of intelligence in an artificial device.
The term was coined by McCarthy in 1956.
There are two ideas in the definition.
1. Intelligence g
2. Artificial device
4
What is AI ?
What is intelligence?
S hi h h i h f ll h Something that characterizes humans from all other
beings?
Criteria to measure intelligence or an absolute g
standard of judgment for intelligence?
5
What is AI?
What is intelligence?
R di i lli h ibili i Regarding intelligence, there are two possibilities:
A system with intelligence is expected to behave as
intelligently as a human.
A system with intelligence is expected to behave in the best
possible 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
Intro to AI
Artificial Intelligence (AI) is an interesting sub-field
of comp ter science that pro ides man of computer science that provides many
contributions to the overall field
Intelligence of machines Intelligence of machines
Study and design of intelligent agents
An intelligent agent is a system that perceives its g g y p
environment and takes actions which maximize its
chances of success.
7
Source:Wikipedia
Intro to AI
Major Areas of AI
D d i i bl l i Deduction, reasoning, problem solving
Knowledge representation
Planning Planning
Learning
Natural language processing
Motion and manipulation
Social intelligence

8
Intro to AI
Tools of AI
S h Search
Logic
Handling uncertainty (Fuzzy systems) Handling uncertainty (Fuzzy systems)
Clustering and classification
Neural networks
Genetic algorithms
Reasoning tools
Data mining Data mining

9
Intro to AI
AI Languages
S h / LISP Scheme / LISP
Functional
Simple knowledge representation (list)
Easy to apply functionality to represented elements Easy to apply functionality to represented elements
Prolog
Logic-based
Facts and rules easily represented Facts and rules easily represented
Built-in search engine
Specialized languages
Rule languages (e.g. CLIPS) g g ( g )
Planning languages (e.g. STRIPS)
10
Intro to AI
Definitions four major j
combinations
Based on thinking or
acting.
Based on activity like
humans or performed in
Systems
that think
like humans
Systems
that think
rationally
humans or performed in
rational way.
like humans rationally
Systems
that act like
Systems
that act that act like
humans
that act
rationally
11
Intro to AI
1. Acting Humanly
T i T Turing Test
Who is Turing?
Inventor of modern computers
Turing Thesis
Algorithms Turing machines Systems that
think like
humans
Systems that
think
rationally humans rationally
Systems that Systems that
act like
humans
act rationally
12
Intro to AI
1. Acting Humanly
Th T i T t h th The Turing Test has three
participants -- two subjects and a
judge. One of the subjects is a
person and the other is a computer. person and the other is a computer.
Both subjects are hidden from the
view of the judge. They communicate
with the judge via text-only channels.
The role of the judge is to determine The role of the judge is to determine
which text channel corresponds to
the human and which corresponds to
the computer. If the judge cannot p j g
determine this, then the computer
passes the test.
13
Intro to AI
An Application of the Turing Test - CAPTCHA:
CAPTCHA CAPTCHA:
Completely Automatic Public Turing tests to tell Computers
and Humans Apart
e.g.:
Display visually distorted words
Ask user to recognize these words
E l f li ti h l h il Example of application: have only humans open email
accounts
14
Intro to AI
An Application of the Turing Test - CAPTCHA:
15
Intro to AI
1. Acting Humanly
N h d T i ! No program has yet passed Turing test!
(Annual Loebner competition & prize.)
A program that succeeded would need to be capable p g p
of:
Natural language processing: To enable it to communicate
successfully in English. y g
Knowledge representation: To store what it knows or hears
Automated reasoning: To use the stored information to
answer questions and to draw new conclusions
Machine learning: To adapt to new circumstances and to
detect and extrapolate patterns.
16
Intro to AI
2. Thinking Humanly
T d d h h i d Try to understand how the mind
works - how do we think?
Two possible routes to find p
answers:
By introspection - we figure it out
ourselves!
By experiment - draw upon
techniques of psychology to conduct
controlled experiments. (Rat in a
box !)
Systems
that think
like humans
Systems that
think
rationally
box.!)
The discipline of cognitive
science: particularly influential in
i i t l l
y
Systems
that act like
humans
Systems that
act rationally
17
vision, natural language
processing, and learning.
Intro to AI
2. Thinking Humanly
Human vs Machine Thinking
Expert systems - AI success story in early 80's.
Human expert's knowledge and experience is passed to a Human expert s knowledge and experience is passed to a
computer program
Rule-based representation of knowledge
Typical domains are: Typical domains are:
Medicine (INTERNIST, MYCIN, . . . )
Geology (PROSPECTOR)
Chemical analysis (DENDRAL) Chemical analysis (DENDRAL)
Configuration of computers (R1)
Thinking humanly works!
18
Intro to AI
2. Thinking Humanly
Human vs Machine Thinking
Computer program playing chess
Human way Human way
Tried by World champion M. Botvinnik (who also was a
programmer)
Poor performance Poor performance
Computer way
Sophisticated search algorithms
V d b Vast databases
Immense computing power
Human world champion beaten!!!
19
Intro to AI
3. Thinking Rationally
L f h h h AI 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).
R lt i id li d i Result is idealised reasoning.
Systems
that think
like humans
Systems
that think
rationallyy
Systems
that act like
humans
Systems
that act
rationally
20
y
Intro to AI
3. Thinking Rationally
L i i h h i ll i Logicist approach theoretically attractive.
Lots of problems:
Transduction: How to map the environment to symbolic p y
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
Intro to AI
4. Acting Rationally
A i i ll i hi ' l i Acting rationally = acting to achieve one's goals, given
one's beliefs.
Design a rational agent approach to AI g g pp
An agent is just something that acts. Computer agents are
expected 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
Adapting to change
Systems that
think like
humans
Systems that
think
rationally
Adapting to change
y
Systems that
act like
humans
Systems that
act rationally
22
Intro to AI
4. Acting Rationally
E h i hif f d i i h i ll b Emphasis shifts from designing theoretically best
decision making procedure to best decision making
procedure possible in circumstances.
Achieving perfect rationality (making the best decision
theoretically possible) is not usually possible, due to
Limited resources
Limited time
Limited computational power
Limited memory Limited memory
Limited or uncertain information about environment
The trick is to do the best with what you've got!
23
Intro to AI
1950
T ring predicted that in abo t fift ears "an a erage Turing predicted that in about fifty years "an average
interrogator will not have more than a 70 percent chance
of making the right identification after five minutes of
questioning".
1957
Newell and Simon predicted that "Within ten years a
computer will be the world's chess champion, unless the
rules bar it from competition " rules bar it from competition.
24
Intro to AI
4. Acting Rationally
D i i l h AI Design a rational agent approach to AI
Rational agent is one that acts so as to achieve the best
outcome or, when there is uncertainty, the best expected
o tcome outcome.
Making correct inferences is sometimes part of being a rational
agent, because one way to act rationally is to reason logically to
the conclusion that a given action will achieve one's goals and g g
then 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
correct thing to do, yet something must still be done. correct thing to do, yet something must still be done.
25
Can Machines Act/Think Intelligently?
Yes, if intelligence is narrowly defined as information
processing processing.
AI has made impressive achievements showing that
tasks initially assumed to require intelligence can be
t t d automated.
But each success of AI seems to push further the limits
of what we consider intelligence.
26
Typical AI Problems
While studying the typical range of tasks that we
might e pect an intelligent entit to perform e might expect an intelligent entity to perform, we
need to consider both common tasks as well as
expert tasks. p
27
Typical AI Problems
Common tasks include
R i i l bj t Recognizing people, objects.
Communicating (through natural language).
Navigating around obstacles on the streets.
These tasks are done matter of factly and routinely by
people and some other animals.
28
Typical AI Problems
Expert tasks include:
M di l di i Medical diagnosis
Mathematical problem solving
Playing games like chess Playing games like chess
29
Typical AI Problems
Computer systems have been able to perform
sophisticated tasks like medical diagnosis sophisticated tasks like medical diagnosis,
performing symbolic integration, proving theorems
and playing chess. p y g
However, on the other hand, it has proved to be
very hard to make computer systems perform
ti t k th t ll h d l t f many routine tasks that all humans and a lot of
animals can do.
Examples of such tasks include navigating our way Examples of such tasks include navigating our way
without running into things, catching prey and avoiding
predators. Humans and animals are also capable of
interpreting complex sensory information
30
interpreting complex sensory information.
Intelligent Behaviour
Some of the tasks and applications that show
intelligent beha io r are intelligent behaviour are:
Perception involving image recognition and computer
vision
Reasoning
Learning
U d t di l i l i t l l Understanding language involving natural language
processing, speech processing
Solving problems
Robotics
31
Approaches to AI
Strong AI
I i b ild hi h l d It aims to build machines that can truly reason and
solve problems. These machines should be self-aware
and their overall intellectual ability needs to be
indistinguishable from that of a human being.
Strong AI maintains that suitably programmed
machines are capable of cognitive mental states. p g
32
Approaches to AI
Weak AI
I d l i h h i f f f It deals with the creation of some form of computer-
based artificial intelligence that cannot truly reason and
solve problems, but can act as if it were intelligent.
Weak AI holds that suitably programmed machines can
simulate human cognition.
33
Approaches to AI
Applied AI
I i d i ll i bl " " It aims to produce commercially viable "smart" systems
For example, a security system that is able to recognise the
faces of people who are permitted to enter a particular
b ilding building.
Applied AI has already enjoyed considerable success.
34
Approaches to AI
Cognitive AI
C d h i b h h 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
Main Areas of AI
Knowledge representation
S h i ll h i ti Search, especially heuristic
search (puzzles, games)
Planning
R i
Robotics
Perception
Reasoning under uncertainty,
including probabilistic
reasoning
Search
Reasoning
Learning
g
Learning
Agent architectures
Knowledge
rep.
Planning
Constraint
satisfaction
Robotics and perception
Natural language processing
Natural
language
...
Expert
36
language
p
Systems
Limitations of AI Limitations of AI
What can AI systems do (limited success) ?
In Computer vision, the systems are capable of face recognition p , y p g
In Robotics, we have been able to make autonomous vehicles.
In Natural language processing, we have systems that are capable
of 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
Planning and sched ling s stems had been emplo ed in sched ling Planning and scheduling systems had been employed in scheduling
experiments with the Hubble Telescope.
The Learning systems are capable of doing text categorization into
about a 1000 topics about a 1000 topics
In Games, AI systems can play at the Grand Master level in chess
(world champion), checkers, etc.
37
Limitations of AI Limitations of AI
What can AI systems NOT do yet?
Understand nat ral lang age rob stl (e g read and Understand natural language robustly (e.g., read and
understand articles in a newspaper)
Surf the web
Interpret an arbitrary visual scene
Learn a natural language
C t t l i d i l ti d i Construct plans in dynamic real-time domains
Exhibit true autonomy and intelligence
38
Foundations of AI
The following disciplines contributed ideas,
ie points and techniq es to AI viewpoints, and techniques to AI.
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 knowledge lead to action? g
39
Foundations of AI
Philosophy (428 B.C .-present)
Aristotle (384 322 B C ) was the first to formulate a precise Aristotle (384-322 B.C.) was the first to formulate a precise
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 initial g y g
premises.
40
Foundations of AI
Philosophy (428 B .C .-present)
Rene Descartes (1596 1650) gave the first clear discussion of Rene Descartes (1596-1650) gave the first clear discussion of
the distinction between mind and matter and of the problems
that arise.
Descartes was also a proponent of dualism. Descartes was also a proponent of dualism.
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
treated as machines treated as machines.
An alternative to dualism is materialism, which holds that the
brain's operation according to the laws of physics constitutes the
mind.
41
Foundations of AI
Philosophy (428 B .C .-present)
Logical positivism also called logical empiricism a Logical positivism, also called logical empiricism, a
philosophical doctrine formulated in Vienna in the 1920s,
according to which scientific knowledge is the only kind of
factual knowledge and all traditional metaphysical doctrines g p y
are to be rejected as meaningless.
42
Source: http://www.britannica.com/EBchecked/topic/346336/logical-positivism
Foundations of AI
Mathematics (c. 800-present)
What are the formal rules to draw valid conclusions? What are the formal rules to draw valid conclusions?
What can be computed?
How do we reason with uncertain information?
43
Foundations of AI
Mathematics (c. 800-present)
Al Khowarazmi a Persian mathematician of the 9th century Al-Khowarazmi, a Persian mathematician of the 9th century,
introduced Arabic numerals and algebra.
George Boole (1815-1864) began mathematical development
and who worked out the details of propositional, or Boolean, and who worked out the details of propositional, or Boolean,
logic.
Gottlob Frege (1848-1925) extended Boole's logic to include
objects and relations, creating the first-order logic (used today
in basic knowledge representation).
Euclid proposed the first nontrivial algorithm for computing
greatest common denominators.
44
Foundations of AI
Mathematics (c. 800-present)
In 1900 David Hilbert (1862 1943) presented a list of 23 In 1900, David Hilbert (1862-1943) presented a list of 23
problems that he correctly predicted would occupy
mathematicians for the bulk of the century.
In 1930, Kurt Godel (1906-1978) , in his incompleteness In 1930, Kurt Godel (1906 1978) , in his incompleteness
theorem showed that in any language expressive enough to
describe theproperties of the natural numbers, there are true
statements that are undecidable in the sense that their truth
cannot be established by any algorithm cannot be established by any algorithm.
45
Foundations of AI
Mathematics (c. 800-present)
Alan Turing (1912 1954) to try to characterize exactly which Alan Turing (1912-1954) to try to characterize exactly which
functions are capable of being computed. The Church-Turing
thesis, which states that the Turing machine is capable of
computing any computable function. Turing also showed that p g y p g
there were some functions that no Turing machine can
compute.
46
Foundations of AI
Economics (1776-present)
How should we make decisions so as to maximize payoff? How should we make decisions so as to maximize payoff?
How should we do this when others may not go along?
How should we do this when the payoff may be far in the
future? future?
47
Foundations of AI
Economics (1776-present)
Lkon Walras (1834 1910) formalized the mathematical Lkon Walras (1834- 1910) formalized the mathematical
treatment of "preferred outcomes7' or utility and was improved
by Frank Ramsey (193 1) and later by John von Neumann and
Oskar Morgenstern. g
Decision theory, which combines probability theory with utility
theory, provides a formal and complete framework for
decisions (economic or otherwise) made under uncertainty
48
Foundations of AI
Neuroscience (1861-present)
How do brains process information? How do brains process information?
49
Foundations of AI
Neuroscience (1861-present)
How do brains process information? How do brains process information?
Neuroscience is the study of the nervous system, particularly
the brain. The exact way in which the brain enables thought is the brain. The exact way in which the brain enables thought is
one of the great mysteries of science.
50
Foundations of AI
Neuroscience (1861-present)
Paul Broca's (1824 1880) study of aphasia (speech deficit) in Paul Broca s (1824-1880) study of aphasia (speech deficit) in
brain-damaged patients in 1861 persuaded the medical
establishment of the existence of localized areas of the brain
responsible for specific cognitive functions. p p g
Hans Berger (1929) invented electroencephalograph (EEG)
for the measurement of intact brain activity.
51
Foundations of AI
Psychology (1879-present)
How do humans and animals think and act? How do humans and animals think and act?
52
Foundations of AI
Psychology (1879-present)
Hermann von Helmholtz (1821 1894) and his student Hermann von Helmholtz (1821-1894) and his student
Wilhelm Wundt (1832-1920) applied the scientific method to
the study of human vision.
Wundt opened the first laboratory of experimental psychology at p y p p y gy
the University of Leipzig.
John Watson (1878-1958) initiated Behaviorism movement
that studies objective measures of the percepts (or stimulus)
given to any animal and its resulting actions (or response) given to any animal and its resulting actions (or response).
Mental constructs such as knowledge, beliefs, goals, and
reasoning steps were dismissed as unscientific "folk psychology."
53
Foundations of AI
Psychology (1879-present)
William James (1842 1910): explained functioning of the William James (1842-1910): explained functioning of the
brain as an information-processing device.
Kenneth Craik (1943) specified the three key steps of a
knowledge-based agent: knowledge based agent:
(1) the stimulus must be translated into an internal
representation,
(2) the representation is manipulated by cognitive processes
t d i i t l t ti d to derive new internal representations, and
(3) these are in turn retranslated back into action.
Anderson 1980: A cognitive theory should be like a computer
program program .
54
Foundations of AI
Computer Engineering (1940 present)
H b ild ffi i t t ? How can we build an efficient computer?
For artificial intelligence to succeed, we need two g ,
things: intelligence and an artifact.
The computer has been the artifact of choice.
55
Foundations of AI
Computer Engineering (1940 present)
Abacus (7000 years old) Abacus (7000 years old)
Pascaline: Mechanical adder & substractor (Pascal; mid
1600s)
Leibniz added multiplication 1694 Leibniz added multiplication, 1694
Analytic Engine: universal computation; never completed
(ideas: addressable memory, stored programs, conditional
jumps)
Charles Babbage (1792-1871), Ada Lovelace
56
Foundations of AI
Computer Engineering (1940 present)
Heath Robinson: First operational electromechanical Heath Robinson: First operational electromechanical
computer built by Alan Turing team in1940, England.
Deciphering German messages.
Z-3: First programmable computer built by Z 3: First programmable computer built by
Konrad Zuse 1941, Germany
ABC: First electronic computer built by John Atanasoff 1940-
42, US 42, US
ENIAC: First general-purpose, electronic, digital computer built
by John Mauchy & John Eckert
57
Birth of AI
Dartmouth 1956 workshop for 2 months
T ifi i l i lli Term artificial intelligence
Fathers of the field introduced
Logic Theorist: program for proving theorems by Logic Theorist: program for proving theorems by
Alan Newell & Herbert Simon
58
Birth of AI
Early Enthusiasm (1952-69)
Cl i d X Claims: computers can do X
General Problem Solver, Newell & Simon
Intentionally solved puzzles in a similar way as humans do. y p y
Geometry Theorem Prover, Herbert Gelernter, 1959
Arthur Samuels learning checkers program, 1952
LISP ti h i Ad i t k M C th 1958 LISP, time sharing, Advice taker: McCarthy 1958
Integration, IQ geometry problems
Blocks world: vision, learning, NLP, planning Blocks world: vision, learning, NLP, planning
Adalines [Widrow & Hoff 1960], perceptron
convergence theorem [Rosenblatt 1962]
59
Birth of AI
A Dose of Reality (1966-74)
Si l i i l i did l Simple syntactic manipulation did not scale
Intractability
Machine evolution did not scale Machine evolution did not scale
Perceptrons book with negative result on
representation capability of 1-layer ANNs [Minsky &
Papert] Papert]
60
Birth of AI
Knowledge-based systems (1969-79)
DENDRAL l l id ifi i DENDRAL: molecule structure identification
[Feigenbaum et al.]
Knowledge intensive
Mycin: medical diagnosis [Feigenbaum, Buchanan,
Shortliffe]
450 rules; knowledge from experts; no domain theory 450 rules; knowledge from experts; no domain theory
Better than junior doctors
Certainty factors
PROSPECTOR: drilling site choice [Duda et al] PROSPECTOR: drilling site choice [Duda et al]
Domain knowledge in NLP
Knowledge representation: logic, frames...
61
Birth of AI
AI becomes an industry (1980-88)
R1 fi f l i l R1: first successful commercial expert system,
configured computer systems at DEC; saved
40M$/year
1988: DEC had 40 expert systems, DuPont 100...
1981: Japans 5th generation project
Software tools for expert systems: Carnegie Group Software tools for expert systems: Carnegie Group,
Inference, Intellicorp, Teknowledge
LISP-specific hardware: LISP Machines Inc, TI,
S b li X Symbolics, Xerox
Industry: few M$ in 1980 -> 2B$ in 1988
62
Birth of AI
Recent events (1987-)
R l ld li ti th th t d i Real-world applications rather than toy domains
Building on existing work
e.g. speech recognition
Ad h f il th d i 1970 Ad hoc, fragile methods in 1970s
Hidden Markov models now
e.g. planning (unified framework helped progress)
Normative system design Normative system design
Belief networks & probabilistic reasoning
Reinforcement learning
Multiagent systems Multiagent systems
63
AI Prehistory
Philosophy Logic, methods of reasoning, mind as physical
system foundations of learning, language,
ti lit rationality
Mathematics Formal representation and proof algorithms,
computation, (un)decidability, (in)tractability,
probability p y
Economics utility, decision theory
Neuroscience physical substrate for mental activity
Psychology phenomena of perception and motor control,
i t l t h i experimental techniques
Computer building fast computers
engineering
Control theorydesign systems that maximize an objective Control theorydesign systems that maximize an objective
function over time
Linguistics knowledge representation, grammar
64
Abridged History of AI
1943 McCulloch & Pitts: Boolean circuit model of brain
1950 Turing's "Computing Machinery and Intelligence" g p g y g
1956 Dartmouth meeting: "Artificial Intelligence" adopted
195269 Look, Ma, no hands!
1950s Early AI programs, including Samuel's checkers
N ll & Si ' L i Th i t program, Newell & Simon's Logic Theorist,
Gelernter's Geometry Engine
1965 Robinson's complete algorithm for logical reasoning
196673 AI discovers computational complexity 1966 73 AI discovers computational complexity
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
IBM Deep Blue:
The IBM computer defeated the reigning world chess The IBM computer defeated the reigning world chess
champion Garry Kasparov in 1997.
66
State of the art
Proof of Robbins Conjecture:
Proved a mathematical conjecture (Robbins conjecture) Proved a mathematical conjecture (Robbins conjecture)
unsolved for decades.
Dr. William McCune at
Argonne Labs Illinois in Argonne Labs, Illinois in
his office with computer.
The " Proof of Robbins
Conjecture" problem is
67
on the screen.
State of the art
Autonomous Control:
The ALVINN computer vision system was trained to steer a car The ALVINN computer vision system was trained to steer a car
to keep it following a lane.
It was placed in CMU's NAVLAB computer-controlled minivan
and was used to navigate across the United States-for 2850 and was used to navigate across the United States for 2850
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
Logistics Planning:
During the 1991 Gulf War US forces deployed an AI logistics During the 1991 Gulf War, US forces deployed an AI logistics
planning and scheduling program that involved up to 50,000
vehicles, cargo, and people.
NASA's on-board autonomous planning program controlled
the scheduling of operations for a spacecraft the scheduling of operations for a spacecraft.
69
State of the art
Language Understanding and Problem Solving:
PROVERB is a computer program that solves crossword PROVERB is a computer program that solves crossword
puzzles better than most humans, using constraints on
possible word fillers, a large database of past puzzles, and a
variety of information sources including dictionaries and online y g
databases such as a list of movies and the actors that appear.
70
AI and Ethics
Ethical Concerns: Robot behavior
How do we want our intelligent systems to behave? How do we want our intelligent systems to behave?
How can we ensure they do so?
Asimovs Three Laws of Robotics:
1 A robot may not injure a human being or through inaction 1. A robot may not injure a human being or, through inaction,
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 such
protection does not conflict with the First or Second Law.
71
AI and Ethics
Ethical Concerns: Human behavior
Is it morally justified to create intelligent systems with these Is it morally justified to create intelligent systems with these
constraints?
As a secondary question, would it be possible to do so?
Should intelligent systems have free will? Can we prevent Should intelligent systems have free will? Can we prevent
them from having free will??
Will intelligent systems have consciousness? (Strong AI)
If they do, will it drive them insane to be constrained by artificial y , y
ethics placed on them by humans?
If intelligent systems develop their own ethics and morality, will
we like what they come up with?
72
Summary
Different people think of AI differently. Two important
questions to ask are: Are you concerned with thinking questions to ask are: Are you concerned with thinking
or behavior? Do you want to model humans or work
from an ideal standard?
I thi b k d t th i th t i t lli i In this book, we adopt the view that intelligence is
concerned mainly with rational action. Ideally, an
intelligent agent takes the best possible action in a
it ti situation.
73
Summary
The history of AI has had cycles of success, misplaced
optimism and resulting cutbacks in enthusiasm and optimism, and resulting cutbacks in enthusiasm and
funding. There have also been cycles of introducing
new creative approaches and systematically refining
the best ones the best ones.
AI has advanced more rapidly in the past decade
because of greater use of the scientific method in
i ti ith d i h experimenting with and comparing approaches.
The subfields of AI have become more integrated, and
AI has found common ground with other disciplines.
74