Chapter One

Introduction to Artificial Intelligence

What is AI?
We call ourselves Homo sapiens-man the wise-because our mental capacities are so importantto
us. For thousands of years, we have tried to understand how we think; that is, howa mere handful of stuff
can perceive, understand, predict, and manipulate a world far largerand more complicated than itself. The
field of artificial intelligence, or AI, goes further still:it attempts not just to understand but also to build
intelligent entities.
A1 is one of the newest sciences. Work started in earnest soon after World War 11, andthe name
itself was coined in 1956. Along with molecular biology, A1 is regularly cited asthe "field I would most
like to be in" by scientists in other disciplines. A student in physics might reasonably feel that all the good
ideas have already been taken by Galileo, Newton,Einstein, and the rest. AI, on the other hand, still has
openings for several full-time Einsteins.A1 currently encompasses a huge variety of subfields, ranging
from general-purpose areas, such as learning and perception to such specific tasks as playing chess,
proving mathematical theorems, writing poetry, and diagnosing diseases. A1 systematizes and automates
intellectual tasks and is therefore potentially relevant to any sphere of human intellectual activity. In this
sense, it is truly a universal field.
We have claimed that A1 is exciting, but we have not said what it is. Definitions of artificial
intelligence according to eight textbooks are shown in the table below. These definitions varyalong two
main dimensions. Roughly, the ones on top are concerned with thought processesand reasoning,
whereas the ones on the bottom address behavior. The definitions on the leftmeasure success in terms of
fidelity to human performance, whereas the ones on the rightmeasure against an ideal concept of
intelligence, which we will call rationality. A system isrational if it does the "right thing," given what it
knows.

Definitions of AI according to eight Textbooks

1
 Historically, all four approaches to A1 have been followed. As one might expect, atension exists
between approaches centered-around humans and approaches centered-aroundrationality.' A human-
centered approach must be an empirical science, involving hypothesisand experimental confirmation. A
rationalist approach involves a combination of mathematicsand engineering. Each group has both
ridiculed and helped the other. Let us look at thefour approaches in more detail.

Acting humanly: The Turing Test approach

TheTuring Test, proposed by Alan Turing (195O), was designed to provide a satisfactory
operational definition of intelligence. Rather than proposing a long and perhaps controversiallist of
qualifications required for intelligence, he suggested a test based on indistinguish ability from undeniably
intelligent entities-human beings. The computer passes the test if a human interrogator, after posing some
written questions, cannot tell whether the written responsescome from a person or not. Programming a
computer to pass the testprovides plenty to work on. The computer would need to possess the following
capabilities:

 Natural language processing to enable it to communicate successfully in English.

 Knowledge representation to store what it knows or learns;
 Automated reasoning to use the stored information to answer questions and to drawnew
conclusions;
 Machine learning to adapt to new circumstances and to detect and extrapolate patterns.

A1 researchers have devoted little effort to passingthe Turing test, believing that it is more important
to study the underlying principles of intelligencethan to duplicate an exemplar. The quest for "artificial
flight" succeeded when theWright brothers and others stopped imitating birds and learned about
aerodynamics. Aeronauticalengineering texts do not define the goal of their field as making "machines
that flyso exactly like pigeons that they can fool even other pigeons."

Thinking humanly: The cognitive modeling approach

If we are going to say that a given program thinks like a human, we must have some way of
determining how humans think. We need to get inside the actual workings of human [Link] are two
ways to do this: through introspection-trying to catch our own thoughts as they go by and through
psychological experiments. Once we have a sufficiently precise theory of the mind, it becomes possible to
express the theory as a computer program. If the program's input/output and timing behaviors match
corresponding human behaviors, that is evidence that some of the program's mechanisms could also be
operating in humans. For example ,Allen Newel1 and Herbert Simon, who developed GPS, the "General
Problem Solver", were not content to have their program solve problems correctly. They were more
concerned with comparing the trace of its reasoning steps to traces of human subjects solving the same
problems. The interdisciplinary field of cognitive science brings together computer models from A1 and
experimental techniques from psychology to try to construct precise and testable theories of the workings
of the human mind.

Thinking rationally: The "laws of thought" approach

The Greek philosopher Aristotle was one of the first to attempt to codify "right thinking," that is,
irrefutable reasoning processes. His syllogisms provided patterns for argument structuresthat always
yielded correct conclusions when given correct premises-for example, "Socratesis a man; all men are
mortal; therefore, Socrates is mortal." These laws of thought weresupposed to govern the operation of the
mind; their study initiated the field called logic.
Logicians in the 19th century developed a precise notation for statements about all kindsof things
in the world and about the relations among them. By 1965, programs existed that could, in principle,
solve any solvable problem described in logical notation. The so-called logicisttradition within artificial
intelligence hopes to buildon such programs to create intelligent [Link] are two main obstacles to

2
this approach. First, it is not easy to take informal knowledge and state it in the formal terms required by
logical notation, particularly when the knowledge is less than 100% certain. Second, there is a big
difference between being able to solve a problem "in principle" and doing so in practice. Even problems
with just a few dozenfacts can exhaust the computational resources of any computer unless it has some
guidanceas to which reasoning steps to try first. Although both of these obstacles apply to any attemptto
build computational reasoning systems, they appeared first in the logicist tradition.

Acting rationally: The rational agent approach

An agent is just something that acts (agent comes from the Latin agere, to do). But computer
agents are expected to have other attributes that distinguish them from mere "programs,"such as operating
under autonomous control, perceiving their environment, persisting over aprolonged time period,
adapting to change, and being capable of taking on another's goals. A rational agent is one that acts so as
to achieve the best outcome or, when there is uncertainty,the best expected outcome.
In the "laws of thought" approach to AI, the emphasis was on correct inferences. Making correct
inferences is sometimes part of being a rational agent, because one way to actrationally is to reason
logically to the conclusion that a given action will achieve one's goalsand then to act on that conclusion.
On the other hand, correct inference is not all of rationality,because there are often situations where there
is no provably correct thing to do, yetsomething must still be done. There are also ways of acting
rationally that cannot be said toinvolve inference. For example, recoiling from a hot stove is a reflex
action that is usuallymore successful than a slower action taken after careful deliberation.
All the skills needed for the Turing Test are there to allow rational actions. Thus, weneed the
ability to represent knowledge and reason \with it because this enables us to reachgood decisions in a
wide variety of situations. We need to be able to generate comprehensiblesentences in natural language
because saying those sentences helps us get by in a complexsociety. We need learning not just for
erudition, but because having a better idea of how theworld works enables us to generate more effective
strategies for dealing with it. We needvisual perception not just because seeing is fun, but to get a better
idea of what an actionmight achieve-for example, being able to see a tasty morsel helps one to move
toward it.
For these reasons, the study of A1 as rational-agent design has at least two [Link], it is
more general than the "laws of thought" approach, because correct inference is justone of several possible
mechanisms for achieving rationality. Second, it is more amenable to scientific development than are
approaches based on human behavior or human thought because the standard of rationality is clearly
defined and completely general. Human behavior,on the other hand, is well-adapted for one specific
environment and is the product, in part,of a complicated and largely unknown evolutionary process that
still is far from producing perfection

3
The Foundation of AI
In this section, a brief history of the disciplines that contributed ideas, viewpoints, and techniques to AI is
summarized. The history is organized around a series of questions , and the theories & findings that
contributed to answer the questions. It should be noted that the impression of the history is not to say that
the disciplines have all been working toward AI as their ultimate fruitition and the stated questions are
their only inquiry.

Philosophy (428 B.C – Present)

 Can formal rules be used to draw valid conclusions?
o Given initial premises, one can generate conclusions mechanically (Aristotle, 384-322
B.C)
o Useful reasoning could actually be carried out by a mechanical artifact (Ramon Lull,
1315)
o Reasoning is like numerical computation, that “we add and subtract in our silent
thoughts” (Hobbes, 1588-1679)
o “Arithmetical machine produces effects which appear nearer to thought than all the
actions of animals” (Pascal, 1642)
 Where does knowledge come from?
o “Nothing is in the understanding; which was not in the senses” (John Locke, 1632-1704)
o General rules are acquired by exposure to repeated associations between their elements
(David Hume, 1711-1776)
 How does knowledge lead to action?
o Actions are justified by a logical connection between goals and knowledge of the action’s
outcome (Aristotle)
o Note- this question is vital to AI, because intelligence requires action as well as
reasoning. Moreover, only by understanding how actions are justified can we understand
how to build an agent whose actions are justifiable (or rational)

Mathematics (c.800-present)
 What are the formal rules to draw valid conclusion?
o George Boole (1815=1864) worked out the details of prepositional or Boolean, logic.
o GattlobFrege (1848-1925) extended Boole’s logic to include objects and relations,
creating the first-order logic that is used today as the most basic knowledge
representation system.

Neuroscience (1861-present)
 How do brain process information?
o The brain was widely recognized as the seat of consciousness (mid. 18 th century)
o The existence of localized area of the brain responsible for specific cognitive function
established (Paul Broca, 1824-1880)
o The measurement of intact brain activity and measurements that correspond to ongoing
cognitive processes by functional magnetic resonance imaging (fMRI) is developed
(Ogawa, 1990)

Psychology (1879-present)
 How do humans and animals think and act?
o The scientific method to the study of human vision was applied, which is the origin of
scientific psychology (Hermann Von Helmholtz, 1821-1894)

4
 o Helmholtz also insisted that perception involved a form of unconscious logical inference.
o The first information-processing models of psychological phenomena was developed
(Donald Broadbent, 1858)

Computer engineering (1840-present)

 How can we build an efficient computer?
o The first operational computer was build (Alan Turing’s group, 1940)
o The first operational programmable computer was invented (KonradZuse, 1941)
o In the half century since then, performance doubles every 18 months or so- with a decade
to go at this rate of increase.

Linguistics (1957-present)
 How does language relate to thought?
o “how a child could understand and make up sentences that he has never heard before”
was explained (Noam Chomsky, 1957)
o Understanding language requires an understanding of the subject matter and context, not
just an understanding of the structure of sentences (1960s)

History of AI
The gestation of artificial intelligence (1943-1955)
The first work that is now generally recognized as A1 was done by Warren McCulloch andWalter
Pitts (1943). They drew on three sources: knowledge of the basic physiology andfunction of neurons in
the brain; a formal analysis of propositional logic due to Russell andWhitehead; and Turing's theory of
computation. They proposed a model of artificial neuronsin which each neuron is characterized as being
"on" or "off," with a switch to "on" occurringin response to stimulation by a sufficient number of
neighboring neurons. The state of aneuron was conceived of as "factually equivalent to a proposition
which proposed its adequatestimulus." They showed, for example, that any computable function could be
computed bysome network of connected neurons, and that all the logical connectives (and, or, not,
etc.)could be implemented by simple net structures. McCulloch and Pitts also suggested thatsuitably
defined networks could learn. Donald Hebb demonstrated a simple updatingrule for modifying the
connection strengths between neurons. His rule, now called Hebbianlearning, remains an influential
model to this day.
There were a number of early examples of work that can be characterized as AI, but itwas Alan
Turing who first articulated a complete vision of A1 in his 1950 article "ComputingMachinery and
Intelligence." Therein, he introduced the Turing test, machine learning,genetic algorithms, and
reinforcement learning.

The birth of artificial intelligence (1956)

Princeton was home to influential figure in AI, John McCarthy. After graduation,McCarthy
moved to Dartmouth College, which was to become the official birthplace of [Link] researchers
from Carnegie Tech, Allen Newell and Herbert Simon, rather stolethe show. Although the others had
ideas and in some cases programs for particular applicationssuch as checkers, Newel1 and Simon already
had a reasoning program, the LogicTheorist (LT), about which Simon claimed, "We have invented a
computer program capableof thinking non-numerically, and thereby solved the venerable mind-body
problem." Soonafter the workshop, the program was able to prove most of the theorems.
The Dartmouth workshop did not lead to any new breakthroughs, but it did introduceall the major
figures to each other. For the next 20 years, the field would be dominated bythese people and their

5
students and colleagues at MIT, CMU, Stanford, and IBM. Perhapsthe longest-lasting thing to come out
of the workshop was an agreement to adopt McCarthy'snew name for the field: artificial intelligence.

A1 becomes an industry (1980-present)

The first successful commercial expert system, R1, began operation at the Digital
EquipmentCorporation (McDermott, 1982). The program helped configure orders for new
computersystems; by 1986, it was saving the company an estimated $40 million a year. By 1988,DEC 7s
A1 group had 40 expert systems deployed, with more on the way. Du Pont had 100in use and 500 in
development, saving an estimated $10 million a year. Nearly every majorU.S. corporation had its own A1
group and was either using or investigating expert systems.

Applications of AI
What can A1 do today? A concise answer is difficult, because there are so many activities inso
many subfields. Here we sample a few applications.

Autonomous planning and scheduling: A hundred million miles from Earth, NASA'sRemote Agent
program became the first on-board autonomous planning program to controlthe scheduling of operations
for a spacecraft Remote Agent generatedplans from high-level goals specified from the ground, and it
monitored the operation of thespacecraft as the plans were executed-detecting, diagnosing, and recovering
from problemsas they occurred.

Game playing: IBM's Deep Blue became the first computer program to defeat theworld champion in a
chess match when it bested Garry Kasparov by a score of 3.5 to 2.5 in an exhibition match. Kasparov said
that he felt a "new kind ofintelligence" across the board from him. Newsweek magazine described the
match as "Thebrain's last stand."

Autonomous control: The ALVINN computer vision system was trained to steer a carto keep it
following a lane. Steer was placed in CMU's NAVLAB computer-controlled minivanand used to navigate
across the United States-for 2850 miles it was in control of steering thevehicle 98% of the time. A human
took over the other 2%, mostly at exit ramps. NAVLABh asvideo cameras that transmit road images to
ALVINN, which then computes the best directionto steer, based on experience from previous training runs.

Diagnosis: Medical diagnosis programs based on probabilistic analysis have been ableto perform at the
level of an expert physician in several areas of medicine. One system was able to describe a case where a
leading expert on lymph-node pathology scoffs at a program's diagnosisof an especially difficult case.
The creators of the program suggest he ask the computerfor an explanation of the diagnosis. The machine
points out the major factors influencing itsdecision and explains the subtle interaction of several of the
symptoms in this case. Eventually,the expert agrees with the program.

Logistics Planning: During the Persian Gulf crisis of 1991, U.S. forces deployed aDynamic Analysis
and Re-planning Tool, DART to do automatedlogistics planning and scheduling for transportation. This
involved up to 50,000 vehicles,cargo, and people at a time, and had to account for starting points,
destinations, routes, andconflict resolution among all parameters. The AI planning techniques allowed a
plan to begenerated in hours that would have taken weeks with older methods. The Defense Advanced
Research Project Agency (DARPA) stated that this single application more than paid backDARPA's 30-
year investment in AI.

6
Robotics: Many surgeons now use robot assistants in microsurgery. HipNavis a system that uses
computer vision techniques to create a three-dimensionalmodel of a patient's internal anatomy and then
uses robotic control to guide the insertion of hip replacement prosthesis.

Language understanding and problem solving: PROVERB is acomputer program that solves
crossword puzzles better than most humans, using constraintson possible word fillers, a large database of
past puzzles, and a variety of information sourcesincluding dictionaries and online databases such as a list
of movies and the actors that appearin them. For example, it determines that the clue "Nice Story" can be
solved by "ETAGE” because its database includes the clue/solution pair "Story in France ETAGE?' and
because itrecognizes that the patterns "Nice X7 and "X in France" often have the same solution.
Theprogram does not know that Nice is a city in France, but it can solve the [Link] are just a few
examples of artificial intelligence systems that exist today. Notmagic or science fiction-but rather science,
engineering, and mathematics.