You are on page 1of 23

An Introduction to

Artificial Intelligence
Sana Ansari
https://groups.yahoo.com/neo/groups/iq
ra_ai/info
sana_ansari_iqra@yahoo.com

What is Artificial Intelligence?

AI
Define intelligence by the properties it exhibits
An ability to deal with new situations
The ability to solve problems
To answer questions
To devise plans, and so on

Humans Vs. Machines


Symbolic calculation

Numeric calculation

Natural language
understanding

Machine language

Not very precise

Precise

Knowledge

Data

Generalise from experience

Cannot generalise

Deal with many inputs

Cannot deal with noise

Definitions
Artificial intelligence is the study of systems
that act in a way that to any observer would
appear to be intelligent
Artificial Intelligence involves using methods
based on the intelligent behavior of humans
and other animals to solve complex problems

Definitions

(Contd.)

The study of computations that makes it possible


to perceive, reason & act
The branch of science which deals with helping
machines to find solutions to complex problems in
a more human like fashion
In short putting human intelligence into machines

Definitions

(Contd.)

AI is concerned with the design of intelligence


into a system
Design of computer systems which can exhibit
intelligent behaviours
The term was coined by John McCarthy in 1956

Goal of AI
Why do we need AI?
The main goal of AI:
To create useful smart programs that are
able to do tasks that would normally require a
human expert

Approaches to AI
The definitions on top are concerned with thought processes whereas the ones on
the bottom address behavior
The definitions on the left measure success in terms of human performance, whereas
the ones on the right measure against an ideal performance measure, called
rationality

Systems that THINK LIKE


HUMANS

Systems that THINK


RATIONALLY

The Cognitive Modelling


Approach

The Laws of Thought Approach

Systems that ACT LIKE


HUMANS

Systems that ACT RATIONALLY


The Rational Agent Approach

The Turing Test Approach

Approaches to AI

(Contd.)

A human-centered approach involves


observations and hypotheses about human
behavior
A rationalists approach involves a combination
of mathematics and engineering

Thinking humanly: The cognitive modeling


approach
A given program thinks like a human, way of determining how
humans think
Get inside the actual workings of human minds
There are three ways to do this:
Introspectiontrying to catch our own thoughts
Psychological experimentsobserving a person in action
Brain imagingobserving the brain in action

Program's inputoutput behavior matches human behavior,


means program's mechanism could also be operating in humans
Newell and Simon, 1961, General Problem Solver: Compares the
reasoning with human subjects
Cognitive science: brings together computer models from AI &
experiments from psychology to construct theories of the
working of the human mind

Thinking humanly

(Contd.)

Cognitive Science: the study of thought, learning, and mental


organization, which draws on aspects of psychology, linguistics,
philosophy, and computer modeling

12

Acting Humanly: The Turing Test Approach


The Turing Test, proposed by Alan Turing (1950),
was designed to provide a satisfactory operational
definition of intelligence.
Human Interrogator
Human response (via computer)
Computer Response
Who is the person? Or Whether the responses
received, come from a person or from a computer
The interrogator has to determine which is the
human & which is the computer

Turing Test

(Contd.)

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 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
patterns

To pass the total Turing Test, the computer will need


Computer Vision: to perceive objects, and
Robotics: to manipulate objects and move about

Thinking Rationally: The "laws of thought"


approach
Aristotle was one of the first to attempt to codify "right
thinking," that is, thinking/ reasoning process
Syllogism (deductive reasoning) provides patterns for
arguments structures that give correct conclusions from
correct premises
For e.g Socrates is a man; all men are mortal, thus,
Socrates is mortal
These laws of thought were supposed to govern the
operation of the mind; their study initiated the field
called logic
Emphasis is on correct inference: reason logically

Laws of thought

(Contd.)

There are two main obstacles


First, it is not easy to take informal knowledge and
state it in the formal terms
Second, there is a big difference between solving a
problem "in principle" and solving it in practice
Problems without guidance exhaust the computational
resources of any computer in just a few hundred facts o
try first.

Acting rationally: The rational agent approach


Rational behaviour means doing the right thing
What is the right thing?
Maximize the achievement towards a goal given some
information
An agent is just something that acts
Computer agents

Operate under autonomous control


Perceive their environment
Adapt to change
Make correct inferences

Acting rationally.

(Contd.)

A rational agent is one that acts so as to achieve the best


outcome or, when there is uncertainty, the best expected
outcome
Correct inference is not all of rationality; in some situations,
there is no provably correct thing to do, but something must
still be done

18

Disciplines in AI
Many disciplines contribute to goal of modelling intelligent
entities
Computer Science
Psychology (Study of the human mind and its functions, Human
reasoning)
Philosophy (Belief, knowledge, and reality)
Linguistics (structure & meaning of language)
Human Biology (How the brain works)
Neurons

History of AI
Event

Year

First electronic computer

1941

McCulloch & Pitts Artificial Neuron

1943

First Neural Network (Minsky & Edmonds)

1951

Birth of AI. Coined by McCarthy. Dartmouth Conference

1956

Eliza

1965

Dendrall (Chemical analysis system)

1967

SHRDLU (robot arm carried out instructions - english)

1971

Prolog

1972

MYCIN (formally termed first expert system)

1980

Backpropagation for neural nets

1985

Data Mining
Deep Blue AI system defeats human world chess champion

1990's
1997

20

Historical success stories


ALVINN (1989)
Autonomous Land Vehicle In a Neural Network
Camera enabled vision
NN is used to decide which way the car should turn or go

Deep Blue (1997)


Chess playing computer by IBM
World Champion Kasparov was beaten by DB

Mars Exploration rover (2004)


Search for the characteristics of a wide range of rocks & soils

Everyday AI

Cars : Self Parking


Speech Recognition: Security systems
Intelligent Games: Chess
Medicine: Clinical Decision Support Systems
Pattern Recognition: Disease diagnosis
Internet: Search Engines

Types of AI
Strong AI
Giving computer program sufficient processing power,
and by providing it with enough intelligence
A computer that can literally think and is conscious
ANN

Weak AI
There is no genuine intelligence, no self-awareness
Modeling intelligent behavior to solve complex problems
For e.g computer playing chess

You might also like