1

Introduction to Artificial Intelligence

SHCHERBAKOV BOGDAN
 What is AI?
2
• The branch of computer science called Artificial Intelligence
is said to have been born at a conference held at
Dartmouth, USA, in 1956

• The scientists attending that conference represented several

different disciplines: mathematics, neurology,
psychology, electrical engineering, etc

• They had one thing in common:

They all were trying to use the recently developed
computers to simulate various aspects of human
intelligence
 What is AI?
3
• Artificial Intelligence may be defined as the branch of
computer science that is concerned with the
automation of intelligent behavior

• An exact definition of intelligence is not easy to formulate

• However, there are some general abilities which are

universally considered as intelligent
 What is AI?
4
According to Douglas Hofstadter these are:

- To respond to situations very flexibly

If the same response is exhibited each time, the behavior is called
mechanical.
To survive in changing environments, one need to exhibit innovative
behavior (e.g. art of begging)

- To make sense out of ambiguous or contradictory messages

We understand such messages because our knowledge and experience
allows us to place them in context.
(e.g. time flies like an arrow, buy this washing powder versus buy that
washing powder)
 What is AI?
5
- To recognize the relative importance of different elements
of a situation
(e.g. quality versus price of a commodity)

- To find similarities between situations despite differences

which may separate them
(e.g. chairs in two different pictures)

- To draw distinctions between situations despite similarities

which may link them
(e.g. differences in two cars)

These abilities are largely due to knowledge and experience,

which allows you to place an information in its wider context
 What is AI?
6
Another definition of intelligence:

It is the ability to
- perceive inter-relationship of facts
- learn and understand from experience
- acquire and retain knowledge
- respond quickly and successfully to a new situation
 What is AI?
7
Turing test was proposed in 1950.

It is a test to decide whether or not a particular machine is

intelligent.

Predicted that by 2000, a machine might have a 30% chance of

fooling a lay person for 5 minutes
 What is AI?
8
- Contact only through monitor and keyboard
- Machine tries to pose as a human
- If the player cannot distinguish between human and machine,
then machine is considered intelligent
 
Revised Turing Test: A human converses with an unseen respondent
and attempts to determine whether it is a man or machine. If the
computer fools you into thinking that it is a human, than that
machine is intelligent
 What is AI?
9
Sometimes it is possible to program computers to carry on shallow
conversations, in limited areas, and thus fool unsuspecting
humans into believing that they are addressing other humans.

Example: Program ELIZA simulating a psychiatrist.

Person: I miss my children
ELIZA: “Why do you miss your children?”
or “ Tell me more about your family”
 
ELIZA is programmed to ask pre-determined questions and parrot
segments of your responses back to you. Hence Turing test may
not be such a good judge of machine intelligence after all
 What is AI?
10
This leads us to the issue of “understanding”

Even though a machine may be exhibiting intelligent behaviour, it

does not “understands” what it is doing

Searle’s Chinese Room Example

Strong and Weak AI

 Major AI Areas
11
1. Expert Systems
An ES is a computer program designed to act as an expert in a
particular domain (area of expertise). It typically includes a sizeable
knowledge base, consisting of facts about the domain and rules for
application to those facts. Medical (e.g. PXDES, MYCIN) and Agriculture
(e.g. AGREX)
2. Natural Language Processing
Goal is to enable people and computers to communicate in ordinary or
natural English.
- Comprehension of natural language: Keyboard
input (e.g. MS Word Processor), speech recognition (e.g. IBM
VoiceType Dictation , BBN corporation: voice activated
browsers, speaker identification for security and Operetta™)
- Generation of natural language.
 Major AI Areas
12
3. Machine learning
Field of study that gives computer the ability to learn without being explicitly
programmed (Arthur Samuel, 1956)

ML learning provides best methods for developing

particular kinds of software, in applications where:
1. Application is too much complex for
people to manually design the algorithm.
For instance, soft- wares for sensor-based
prediction tasks such as speech
recognition and computer vision.
 Major AI Areas
13
3. Machine learning
Field of study that gives computer the ability to learn without being explicitly
programmed (Arthur Samuel, 1956)

ML learning provides best methods for developing

particular kinds of software, in applications where:
2. Applications require that the software
customize to its operational environment after it
is fielded. For example, speech recognition
system that customize to the user who purchase
the software or recommenders
 Major AI Areas
14
4. Robotics and Computer Vision
Factory automation
Autonomous vehicles
Robots: Electromechanical devices programmed to perform
manual tasks. Not all robots are intelligent. Some are pre-
programmed by conventional techniques and are dumb. An
intelligent robot usually includes some kind of sensory apparatus
that allows it to respond to changes in its environment.
Computer Vision: it is field that include methods for
acquiring, processing, analyzing and understanding images in order
to produce numerical and symbolic information. For example,
medical image processing is an application of computer vision