Overview of Artificial

Intelligence
DEFINITION
NEED AND IMPORTANCE
HISTORY OF AI
TYPES OF AI
BRANCHES OF AI
FIELDS RELATED TO AI
DEVELOPMENTS/
APPLICATIONS OF AI
AI PROGRAMMING LANGUAGES
 Definition

 Is the study of how to make

computers behave intelligently.
 Intelligent computers which can
proceed without human
intervention when faced with
unforeseen situations.
 Especially for non-linear problem
solving, where equations are not
defined.
 Possess ability to learn, simulate,
reason.
 AI is science or engineering?

 AI as science:
AI is linked with biology.
Humans exhibit intelligent behavior.
AI seeks to understand the computations
required for intelligent behavior and to
produce computer systems that exhibit intelligence.
Human concepts studied by AI are perception,
motor control, communication using human
languages, reasoning, planning, learning,
and memory.
 AI as Engineering

 Intelligence involves mechanisms, and

AI research has discovered how to
make computers carry out some of them
and not others.
 If doing a task requires only
mechanisms that are well understood
today, computer programs can give very
impressive performances on these tasks.
 Building systems that behave
intelligently in order to solve
practical problems.
 Design and construct a piece of
software.
 What is Intelligent behavior?

 Intelligent behavior
 Learn from experience
 Apply knowledge acquired from experience
 Handle complex situations
 Solve problems when important information is missing
 Determine what is important
 React quickly and correctly to a new situation
 Understand visual images
 Process and manipulate symbols
 Be creative and imaginative
 Use heuristics (thumb rules)
 Need for AI….
 Tasks such as „numerical computations‟, „storing
huge information‟, „repetitive operations‟ are
considered „mindless‟ or „mechanical‟, which the
computers are already good at.
 Those tasks that are very simple for a human beings such
as „understanding the meaning of a simple
sentence‟ or „recognizing somebody‟s face or a
picture‟ are difficult for a computer to do, unless
programmed for it.
 Human beings prefer to have intelligent computers who
would do the tasks that usually a human is known to do
well.
 This drives for need for AI systems (intelligent
computers)
 Challenges in AI….Its importance…

LOOK AT THE
EXAMPLE:
 Sentence 1 and Sentence 2:
 In sentence 1, who
 „I stood on the table to paint the is „It‟ and where
window. It broke and I fell through
did you fall?
onto the flowered bed below‟

 „I stood on the table to paint the  In sentence 2, who

window. It broke and I fell through is „It‟ and where
onto the carpet below‟ did you fall?
 Contd……….
 Who is „It‟? Is it table or window that broke….It could be
anything.

 However, naturally, when a human being tries to

understand, he/she might think that it was window in first
case, because there are usually flower-beds on the other side
of the window.
 In second sentence, it might be understood as the table,
because you fell on the carpet, which is usually supposed to be
under the table.

 It proves that the analysis of „natural language‟ involves the

common sense knowledge of the world, which is very
difficult and almost impossible to devise rules to cover every
possible situation within a computerized system.
 Contd…

 It will be difficult task to write a computer program

to interpret natural language statements and store
the information required, as sets of statements in
logic, it will be difficult task to ensure that it always
interprets „it‟ to mean the thing that a human would
naturally interpret it to mean………..
 Typical AI problems

 Intelligent entities (or “agents”) need to be able to do

both “mundane” and “expert” tasks:
 Mundane tasks - consider going shopping:
 Planning a route, and sequence of shops to visit!
 Recognising (through vision) buses, people.
 Communicating (through natural language).
 Navigating round obstacles on the street, and
manipulating objects for purchase
 Typical AI problems

 Expert tasks are things like:

 medical diagnosis.

 equipment repair.
Typical AI Problem
Issues involved:
Sensor: Camera
Understanding the
image
Finding the solution
Actuator:
 Based on the solution
 Moving arms to slide tiles
 Advantages of AI

 Smarter artificial intelligence promises to replace human jobs,

freeing people for other pursuits by automating manufacturing
and transportations.

 Self-modifying, self-writing, and learning software relieves

programmers of the burdensome task of specifying the whole of a
program’s functionality—now we can just create the framework and
have the program itself fill in the rest
(example: real-time strategy game artificial intelligence run by a
neural network that acts based on experience instead of an explicit
decision tree).

 Self-replicating applications can make deployment easier and less

resource-intensive.

 AI can see relationships in enormous or diverse bodies of data that a

human could not.
Example
 History of AI

5th century BC
Aristotle invents syllogistic logic, the first formal
deductive reasoning system.

16th century AD
Rabbi Loew supposedly invents the Golem, an
artificial man made out of clay
 Contd…

17th century
Descartes proposes animals are machines and
founds a scientific paradigm that will dominate for
250 years.
Pascal creates the first mechanical calculator in 1642

18th century
Wolfgang von Kempelen “invents” fake chess-playing
machine, The Turk.
 History of AI…

19th century
George Boole creates a binary
algebra to represent “laws of
thought”

Charles Babbage and Lady

Lovelace develop sophisticated
programmable mechanical
computers, precursor to modern
electronic computers.

The picture is of a model built in the late 1800s by Babbage‟s son from Babbage‟s drawings.
 History of AI

20th century
Karel Kapek writes “Rossum‟s Universal Robots”,
coining the English word “robot”

Warren McCulloch and Walter Pitts lay partial

groundwork for neural networks

Turing writes “Computing Machinery and

Intelligence” – proposal of Turing test
 History of AI

1956: John McCarthy coins phrase

“artificial intelligence”. Made Lisp
Programming language

1952-62: Arthur Samuel writes the

first AI game program to challenge
a world champion, in part due to
learning.

1950‟s-60‟s: Masterman et. al at

Cambridge create semantic nets
that do machine translation.
 History of AI

1961: James Slagle writes first symbolic integrator,

SAINT, to solve calculus problems.

1963: Thomas Evan‟s writes ANALOGY, which solves

analogy problems like the ones on IQ tests.

1965: J. A. Robinson invents Resolution Method

using formal logic as its representation language.
 History of AI

1965: Joseph Weizenbaum at MIT, creates ELIZA,

one of the earliest “chatterbots”

1967: Feigenbaum et. al create Dendral, the first

useful knowledge-based agent that interpreted
mass spectrographs.

1969: Shakey the robot combines movement,

perception and problem solving.
 History of AI

1971: Terry Winograd demonstrates a program that

can understand English commands in the word of
blocks.

1972: Alain Colmerauer writes Prolog

1974: Ted Shortliffe creates MYCIN, the first expert

system which showed the effectiveness of rule-based
knowledge representation for medical diagnosis.
 History of AI

1978: Herb Simon wins Nobel Prize for theory of

bounded rationality

1983: James Allen creates Interval Calculus as a

formal representation for events in time.

1980‟s: Backpropagation (invented 1974)

rediscovered and sees wide use in neural networks
 History of AI
1985: ALVINN, “an autonomous land
vehicle in a neural network”
navigates across the country (2800
miles).

Early 1990‟s: Gerry Tesauro creates

TD-Gammon, a learning
backgammon agent that vies with
championship players In 1997, Deep Blue beat
Gary Kasparov.
1997: Deep Blue defeats Garry
Kasparov
Robotics in game playing and surgery
marks the splendid achievements in
AI based research.
Seoul Robotic Football Game (2001)
was an achievement.
Post –Cartesian: Modern times:
Robopets
 Turing Test
 Alan Turing's 1950 article Computing
Machinery and Intelligence discussed
conditions for considering a machine
to be intelligent.

 Alan Turing was a British scientist.

 Intelligence is defined as ability to

achieve human level performance in
all cognitive tests sufficient to fool a
human interrogator.

 The test was devised in response to the

question “Can a computer think?”

 Result was positive if the interrogator

cannot tell whether the responses are
coming from a machine or human.
Alan M. Turing, (1912 - 1954)
 The Turing Test

 One person sits at a

computer and types the
questions.
 The computer is
connected to two other
hidden computers.
 At one computer, human
reads and answers the
questions.
 At the other end, the
computer with no human
aid runs and answers the
questions.
 Turing Machine

 A Turing Machine is a mathematical

model of a digital computer running a
program.
 Alan Turing hypothesized that a TM
can perform any arithmetic /
logical task that we may specify. TM
can perform tasks such as solving
equations.
 There is one TM that can simulate all
other TMs. (Called the Universal
Turing Machine, UTM)
Another AI experiment…Chinese experiment

 Philosopher Searle‟s experiment:

 An English man who does not understand Chinese is
put into a room and made to undergo a thought
experiment.
 The instructions are made available in English to
manipulate and combine Chinese characters.
 A Chinese story is given within the room and the set
of questions are given at end to answer.
 By following the instructions and matching the
characters on basis of their appearance, the English
man can answer the questions given to him.
 Contd…

 The result……………
 Searle showed that in principle, people can behave as
if they are fluent in a foreign language (such as
Chinese) and while actually, not being able to
understand a word of it.
- AI became an industry in 1980s
* Commercial expert system, R1, began operation at Digital
Equipment
Corporation in 1982.
* Japan proposed to develop 5th generation computer (intelligent
computer) project (10 years plan) in 1986.
http://www.japanlaw.com/lawletter/july86/ekf.htm
News
The 8 participants companies are:
Hitachi, NEC, Fujitsu, Toshiba,
Japanese efforts to develop Sharp, Oki Electric, Mitsubishi
artificial intelligence the use of Electronic and Matsushita Electric
computers to draw inferences and Industrial. Mitsubishi Electric in
make judgments -- is gaining May 1986 became the first firm to
momentum as 19 major commercialize a product
companies have recently developed under the 5th
established the Artificial Generation Computer Project
Intelligence Joint Research which aims at developing AI,
Society and the Artificial software development.
Intelligence Center. 31
Robots

Sony says next decade the age of the robot

Asimo in New York as a

stockbroker
http://www.honda.co.jp/ASIMO/event/report_08.ht
Robots--working for Japan's future?
ml

That is one goal of the Japanese government's $37.7 million

Humanoid Robotics Project (HRP), which aims to market within a
few years robots that can operate power shovels, assist
construction workers and care for the elderly. In the process, a
new multibillion-dollar Japanese industry could be born.
Robot competitions/exhibitions around the world
K'NEX K-bot World Championships
"ROBODEX2002“, Pacifico Yokohama Exhibition Hall,
1-1-1, Minatomorai, Nishi-ku, Yokohama-city, Kanagawa, Japan
32
33
 Artificial Life and ethic considerations..

Artificial Life: Human made systems that possess

some of the key properties of human life.

Artificial life may be viewed as more expendable

than human life, so AI will be used as cheap
labor, or perhaps slaves, thus increasing
profits for corporations.

We do have to take responsibility for our creations,

so if the risks associated with creating a form of AI
are too great, then we should not pursue that
development.
What rules should we use as categorical
imperatives?

 AI should contain a set of rules that most people

share, like “do not kill, unless in self-defense” or “do not
lie, unless the suffering caused by honesty is large”.

 What preprogrammed “virtues” should computers

have to allow them to be morally right? Can virtues
make an AI entity behave morally at all?

Wisdom, compassion, courage, strength, obedience,

carefulness…?
 Developments in AI

 Currently, no computers exhibit full artificial

intelligence (that is, are able to simulate human
behavior).
 The greatest advances have occurred in the field of
games playing.
 The best computer chess programs are now
capable of beating humans.
 In May, 1997, an IBM super-computer called Deep
Blue defeated world chess champion Gary
Kasparov in a chess match.
 Developments in AI

 In the area of , computers are now widely used in

assembly plants, but they are capable only of very
limited tasks.

 Robots have great difficulty identifying objects based

on appearance or feel, and they still move and handle
objects clumsily.

 Natural-language processing offers the greatest

potential rewards because it would allow people to
interact with computers without needing any
specialized knowledge. You could simply walk up to a
computer and talk to it.
 Developments in AI

 Programming computers to understand natural

languages has proved to be more difficult than
originally thought.
 Some rudimentary translation systems that translate
from one human language to another are in existence,
but they are not nearly as good as human
translators.
 There are also voice recognition systems that can
convert spoken sounds into written words, but they do
not understand what they are writing; they simply take
dictation.
 Even these systems are quite limited – The user must
speak slowly and distinctly.
 Developments in AI

 In the early 1980s, expert systems were believed to

represent the future of artificial intelligence and of
computers in general.
 To date, however, they have not lived up to
expectations.
 Many expert systems help human experts in such fields
as medicine and engineering, but they are very
expensive to produce and are helpful only in special
situations.
 Today, the hottest area of artificial intelligence is
neural networks, which are proving successful in a
number of disciplines such as voice recognition and
natural-language processing.
 In speech recognition….

 A process of converting a speech signal to a sequence of

words;
 Uses:
 Voice dialing: ex: call home.
 Call routing: Collect call,
 Data entry: credit card number entry etc.
 Speaker recognition: [security]
The spoken language interface PEGASUS in the
American Airlines‟ EAASY SABRE reservation system,
allows users to obtain flight information and make
reservations over the telephone.
 John Searle made a distinction….

 Strong AI: Hard AI

A computer running a piece of software and
performing a particular task is intelligent.
 Machine trying to almost supersede human
intelligence fully.
 The machine does tasks which humans do.
 It can apply wide range of background knowledge
and has some degree of self-consciousness.
 The machine looks for best alternative (search) and
learns from experience (ANNs).
Thus the machines can almost become
indistinguishable from humans.
 And Weak AI

 Weak AI: Soft AI/

Intelligence is something that only biological systems
possess and everything else that machine does is called
simulation.
 The actions are fixed and predetermined.
 In order to improve understanding of reflexes and Context
Awareness can be added
Actions affected by the knowledge of the environment.
Context Information.
It does not exhibit full range of human cognitive abilities.
It is merely intelligent and a specific problem-solver.
Follows a complex set of rules.
Eg: Some Games.
 Goals of AI

 There are 4 popular possible goals of AI

 Systems that can (a) think and (b) act like humans.
 Systems that can (a) think and (b) act rationally.

 This leads to following approaches:

 (1). Think like humans --- cognitive approach
 (2). Act like humans --- Turing test.
 (3). Think rationally --- Laws of Thought approach
 (4). Act rationally --- rational
 (1). Cognitive approach

 It is an effort to make machines think.

 Focus is not on I/O or behavior but on reasoning
process.
 Results are obtained by computational model.
 The goal here is not only to produce human like
behavior but to produce a series of reasoning
steps [as used by a human expert]
 (2). Act like humans

 Making machines do things, that at the moment are

better done by humans.
 The focus is on the action and not mere behavior.
 Example is the „Turing Test‟.

 Practical example: Robot

 (3). Laws of Thought – think rationally

 The study of mental faculties through the use of

computational models make it possible to perceive,
reason and act.
 Focus is on inference mechanisms that are
provably correct and guarantee an optimal solution.
 Example: „Socretes is a man, men are mortal.
Therefore Socretes is mortal.‟
 The goal is to formalize reasoning, as a system of
logical rules and procedures for inference.
 (4). Act Rationally – Rational agent approach

 Tries to explain and emulate intelligent human

behavior in terms of computational processes, that
it is concerned with automation of intelligence.
 Focus is on creating systems that act sufficiently if
not optimally in all situations.
 It is passable to have imperfect reasoning, if the job
gets done.
 Major branches of AI

 Perceptive system
 A system that approximates the way a
human sees, hears, and feels objects
 Vision system
 Capture, store, and manipulate visual
images and pictures
 Robotics
 Mechanical and computer devices that
perform tedious tasks with high precision
 Expert system
 Stores knowledge and makes inferences
 Major branches of AI (CONTD…)

 Learning system
 Computer changes how it functions or reacts to situations based
on feedback
 Natural language processing
 Computers understand and react to statements and commands
made in a “natural” language, such as English
 Neural network
 Computer system that can act like or simulate the functioning of
the human brain
 Pattern Recognition
 Matching and recognition of faces, images etc.
 Schematic Representation
Artificial
intelligence

Vision Learning
systems systems

Robotics
Expert systems

Neural networks
Natural language
processing
Related Fields to AI
AI inherited many ideas, viewpoints and techniques from
other disciplines.

To investigate
human mind Theories of reasoning
and learning

AI

Linguistic
Mathematics
The meaning and Theories of logic
structure of language probability, decision
CS making and computation

Make AI a reality
52
 Related branches…

 Philosophy: logic, brain working, reasoning.

 Mathematics: What can and cannot be computed.
 Psychology: How do humans think? Processes
underlying intelligence.
 Computer Science: How can we build efficient
computer/machine/software.
 Biology: The inner structure of human brain.
 Mechanics: Making robotic system, arms, controls
etc.
 Electronics: Developing circuits and systems
 Wide spread applications of AI
 Business Engineering
• Credit rating and risk • Machinery defect diagnosis
assessment • Signal processing
• Insurance risk evaluation • Character recognition
• Fraud detection • Process supervision
• Insider dealing detection • Process fault analysis
• Marketing analysis • Speech recognition
• Mailshot profiling • Machine vision
• Speech recognition
• Signature verification
• Radar signal classification
• Inventory control
• Transducer linerisation
• Prediction of share and
• Colour discrimination
commodity prices
• Robot control and navigation
• Prediction of economic
indicators • Process control
• Aircraft landing control
• Insider dealing detection
• Car active suspension control
• Marketing analysis
• Printed Circuit auto routing
• Mailshot profiling
• Integrated circuit layout
• Signature verification • Image compression
• Inventory control
 Contd.
 Security
• Face recognition
• Speaker verification
• Fingerprint analysis

 Medicine
• General diagnosis
• Detection of heart defects
• Medical imaging and image
processing
 Contd..

 Science
• Recognising genes
• Botanical classification
• Bacteria identification
• Prediction of the performance of
drugs from the molecular structure
• Weather prediction
• Sunspot prediction
 Forecasting
• Future sales
• Production Requirements
• Market Performance
• Economic Indicators
• Energy Requirements
• Time Based Variables
 Many More

 Many More………………
 AI Languages

 Consider an English sentence like:

 “The boy saw a girl with a telescope”
 Natural languages exhibit ambiguity
 Not only does ambiguity make it difficult for us to
understand what is the intended meaning of certain
phrases and sentences but also makes it very difficult
to make inferences too.
 There are several programming languages that are
known as AI languages because they are used almost
exclusively for AI applications.
 These help to represent logic in formalized manner.
 The two most common are LISP and Prolog.
 Getting Started with Prolog

 Prolog is a language based on first order predicate

logic.
 We can assert some facts and some rules, then ask
questions to find out what is true.
 Facts:
likes(john, mary).
tall(john).
tall(sue).
short(fred).
teaches(alison, artificialIntelligence).

 Note: lower case letters, full stop at end.

 Prolog

 Rules:

likes(fred, X) :- tall(X).
examines(Person, Course) :- teaches(Person, Course).

 John likes someone if that someone is tall.

 A person examines a course if they teach that course.
 NOTE: “:-” used to mean IF. Meant to look a bit like a backwards
arrow
 NOTE: Use of capitals (or words starting with capitals) for
variables.

 Program is made of Rules & Facts.