Artificial Intelligence

Dr. Rasha Orban

Introduction
Chapter 1
 ?What is AI
 A broad field that means different things to
different people
 Defining “artificial” is easy but no broad
consensus in precise, concrete terms for
“intelligence”:
 exclusive province of human being?
 natural phenomenon exhibited by living
organisms?
 an arbitrarily specified set of abilities
3
 What is AI? contd.
 Various other definitions of AI are possible ranging
from philosophical to engineering perspectives
 Machines with minds…
 Study of design of intelligent agents…

 A good working definition: Study of how to make

computers do tasks at which currently humans are
better (Rich and Knight, 1991)
 AI is a study of duplicating human faculties like
creativity, self-improvement and language use
 .What is AI? contd

 AI is the science and engineering of making

intelligent machines, especially intelligent
computer programs. It is related to the
similar task of using computers to
understand human intelligence, but AI does
not have to confine itself to methods that
are biologically observable.
 .What is AI? contd

 “Artificial” Intelligence may seem to

mean that the intelligence is not real or
that it just simulates natural intelligence.
 .What is AI? contd
 “Intelligence” is the computational part of
the ability to achieve goals in the world.
 Varying kinds and degrees of intelligence
occur in people, many animals and some
machines.
 ?What is NOT AI
 Word processing software
 Graphics design using computers
 Entering or searching for an entry in a database
 Computer networks, Internet protocols
 Computer security
 Design of a processor

These tasks do not require the computer being

used to be intelligent.
 What Makes Humans
?Intelligent
 Knowledge
 Reasoning

 Learning

 Language and Perception

All these are reflected in the major sub-areas

of AI
 Symbolic vs. Connectionist
Approaches
 AI research follows two distinct, and to
some extent competing, methods:
 the symbolic (or “top-down”) approach.
 the connectionist (or “bottom-up”)
approach.
 Symbolic vs. Connectionist
Approaches contd.
 The top-down approach seeks to replicate
intelligence by analyzing cognition
independent of the biological structure of the
brain, in terms of the processing of symbols—
whence the symbolic label.

 The bottom-up approach, on the other hand,

involves creating artificial neural networks in
imitation of the brain’s structure—whence
the connectionist label.
 Symbolic vs. Connectionist
Approaches contd.
 To illustrate the difference between these
approaches:
 Consider the task of building a system,
equipped with an optical scanner, that
recognizes the letters of the alphabet.
 Symbolic vs. Connectionist
Approaches contd.
 A bottom-up approach typically involves
training an artificial neural network by
presenting letters to it one by one,
gradually improving performance by
“tuning” the network. (Tuning adjusts
the responsiveness of different neural
pathways to different stimuli.)
 Symbolic vs. Connectionist
Approaches contd.
 In contrast, a top-down approach
typically involves writing a computer
program that compares each letter with
geometric descriptions.
 Simply put, neural activities are the
basis of the bottom-up approach, while
symbolic descriptions are the basis of the
top-down approach.
 Intelligent Behavior
(or stuff people are good at)
• Problem Solving
• Learning
• Planning
• Perception
• Language Processing
• Collecting Stuff
• Independent Action
 ?What is AI
Views of AI fall into four categories:

Thinking humanly Thinking rationally

Acting humanly Acting rationally

The textbook advocates "acting rationally"




 .What is AI? contd
“The exciting new effort to “The study of mental faculties
make computers thinks … through the use of computational
machine
Systemswiththat
minds, in like
think the full models”
Systems that think rationally
and literal sense”
humans (Charniak et al. 1985)
(Haugeland 1985)
“The art of creating machines “A field of study that seeks to
that perform functions that explain and emulate intelligent
Systems that act like
require intelligence when behavior
Systemsin terms of rationally
that act
humans
performed by people” computational processes”
(Kurzweil, 1990) (Schalkol, 1990)
 What is an Intelligent System?
•System that:
•does everything that I want it to do &
nothing that I don’t
•Human-like reasoning
•recover from failure
•Learns from its mistakes
•Adapts to its environment
•Makes decision about appropriateness of
actions
?What are Intelligent Systems
)?What is AI(

Turing test Logic

Systems that think rationally Systems that think like
humans

Systems that act rationally Systems that act like

humans

Cognitive
Agents
science
 Acting humanly: Turing Test
 Turing (1950) "Computing machinery and
intelligence":
 "Can machines think?"  "Can machines
behave intelligently?"
 Operational test for intelligent behavior: the
Imitation Game


 Acting humanly: Turing Test
 Predicted that by 2000, a machine might
have a 30% chance of fooling a lay person
for 5 minutes
 Suggested major components of AI:
 Knowledge;
 Reasoning;

 Language Understanding;

 Learning.


 ?What tasks require AI

 “AI is the science and engineering of

making intelligent machines which can
perform tasks that require intelligence when
performed by humans …”
 ?What tasks require AI
 Solving a differential equation
 Brain surgery

 Inventing stuff

 Playing Jeopardy

 Playing Wheel of Fortune

 ?What tasks require AI
 What about walking?

 What about grabbing stuff?

 What about pulling your hand away from
fire?
 What about watching TV?

 What about day dreaming?

 Thinking humanly: cognitive
modeling
 1960s "cognitive revolution": information-
processing psychology
 Requires scientific theories of internal
activities of the brain

 Thinking humanly: cognitive
modeling
 -- How to validate? Requires
1) Predicting and testing behavior of human
subjects (top-down)
or
2) Direct identification from neurological data
(bottom-up)
 Both approaches (roughly, Cognitive
Science and Cognitive Neuroscience) are
now distinct from AI

 Thinking rationally: "laws of
"thought
 Aristotle: what are correct arguments/thought
processes?
 Several Greek schools developed various forms
of logic: notation and rules of derivation for
thoughts; may or may not have proceeded to the
idea of mechanization



 Thinking rationally: "laws of
"thought
 Direct line through mathematics and
philosophy to modern AI
 Problems:
1. Not all intelligent behavior is mediated by
logical deliberation
2. What is the purpose o f thinking?
What thoughts should I have?
3.


Acting rationally: rational agent
 Rational behavior: doing the right thing
 The right thing: that which is expected to
maximize goal achievement, given the
available information

 Doesn't necessarily involve thinking – e.g.,

blinking reflex – but thinking should be in
the service of rational action



 Rational Agents

 An agent is an entity that perceives

and acts

 This course is about designing

rational agents

 Rational agents
 For any given class of environments and tasks,
we seek the agent (or class of agents) with the
best performance

 Caveat: computational limitations make

perfect rationality unachievable
 design best program for given machine resources

 What is Artificial Intelligence ?
 Goals of AI:
 AI began as an attempt to understand the
nature of intelligence, but it has grown into a
scientific and technological field affecting
many aspects of commerce and society. The
main goals of AI are:
 Engineering: solve real-world problems
using knowledge and reasoning. AI can
help us solve difficult, real-world problems,
creating new opportunities in business,
engineering, and many other application
areas
 What is Artificial Intelligence ?
Goals of AI (cont’d)
 Scientific: use computers as a platform for
studying intelligence itself. Scientists design
theories hypothesizing aspects of intelligence
then they can implement these theories on a
computer.
 Even as AI Technology becomes integrated into the
fabric of everyday life. AI researchers remain
focused on the grand challenges of automating
intelligence.
 AI prehistory
Logic, methods of reasoning, mind as physical
system foundations of learning, language,
Philosophy
rationality
Formal representation and proof algorithms,
computation, (un)decidability, (in)tractability,
Mathematic
probability s
Utility, decision theory
Economics

Physical substrate for mental activity

Neuroscien
ce
 AI prehistory
Phenomena of perception and motor control,
experimental techniques
Psycholog
y
Building fast computers
Computer
engineerin
g
Design systems that maximize an objective function
over time
Control
theory
Knowledge representation, grammar
Linguistics
 Abridged history of AI
McCulloch & Pitts: Boolean circuit model of brain • 1943
Turing's "Computing Machinery and Intelligence" • 1950

Dartmouth meeting: "Artificial Intelligence" adopted • 1956

Look, Ma, no hands! • 1952—69
Early AI programs, including Samuel's checkers program, • 1950s
Newell & Simon's Logic Theorist, Gelernter's Geometry
Engine
Robinson's complete algorithm for logical reasoning • 1965
 Abridged history of AI
AI discovers computational complexity Neural network • 1966—73
research almost disappears
Early development of knowledge-based systems • 1969—79
AI becomes an industry • 1980--
Neural networks return to popularity • 1986--
AI becomes a science • 1987--
The emergence of intelligent agents • 1995--
 AI State of The Art
 Have the following been achieved by AI?
 World-class chess playing
 Playing table tennis
 Cross-country driving
 Solving mathematical problems
 Discover and prove mathematical theories
 Engage in a meaningful conversation
 Understand spoken language
 Observe and understand human emotions
 Express emotions
…
 How to Make Computers
?Intelligent
 Should we model human intelligence
 A good idea but difficult to model
 Human brain is different from computer
processor
 Humans are good at remembering and recognizing
patterns; computers are good at crunching numbers
 A compromising approach: Model human
intelligence as much as possible but also
utilize computer’s ability to crunch numbers
 Airplanes have wings like birds but they don’t
flap them, instead they use engine technology
 Natural vs Machine Intelligence
 Intelligence is not an unique and unshared
capability of human. It is more an open
collection of attributes than it is a single well-
defined entity

 Humans embody many aspects of intelligence

while animals typically embody a smaller number
of intelligent characteristics, and usually at a
much lower level

40
 Natural vs Machine Intelligence

 The advent of digital computers made possible

credible attempts to fulfill the AI dreams

 Computer based intelligence must be specialized

to very restricted domains to be at all comparable
to human performance

41
 Human vs Machine Intelligence
 The two hemispheres of the human brain deal with
problems in two distinct paradigms:
 sequential (or logical) approach that considers only a small
portion of the available data at a time
 parallel (or gestalt) processing looks at data on a global basis

Many tasks which we might Complex

reasonably think require intelligence Arithmetic
are performed by computers without
even thinking
Other tasks that people do without
Recognizing
thinking are extremely difficult to
a Face
automate 42
 AI Problems

AI is concerned with automating both

mundane and expert tasks.
Experts tasks (require specialised skills and training)
include :
 Medical diagnosis
 Equipment repair
 Computer configuration
 Financial planning
43
 AI Problems
Mundane tasks correspond to the following AI problems areas:
 Planning : The ability to decide on a good sequence of
actions to achieve our goals

 Vision : The ability to make sense of what we see

 Robotics: The ability to move and act in the world, possibly
responding to new perceptions

 Natural Language: The ability to communicate with others in

any human language
Mundane tasks are generally much harder to
automate 44
 AI Terminologies
 Strong AI
 machines can actually think intelligently

 Weak AI
 machine can possibly act intelligently

 AIMA: “Most AI researchers take the

weak hypothesis for granted, and don’t
care about the strong AI hypothesis”
45
 ?Applications of AI

Search engines

Science

Medicine/
Diagnosis
Labor
What else?
Appliances 46
 Homework 1 (Due next week)

Should we replace people with

machines? Why?