Lecture1-fundamentals.pptx

ICS 3211: ARTIFICIAL
INTELLIGENCE
LECTURE 1:
INTRODUCTION

Goals of this Course
 This class is a broad introduction to artificial
intelligence (AI)
 AI is a very broad field with many subareas
 We will cover many of the primary concepts/ideas
 But in 10 weeks we can’t cover everything

Why taking ICS 3211could change
your life…..
 As we begin the new millennium
 science and technology are changing rapidly
 “old” sciences such as physics are relatively well-understood
 computers are ubiquitous
 Grand Challenges in Science and Technology
 understanding the brain
 reasoning, cognition, creativity
 creating intelligent machines
 is this possible?
 what are the technical and philosophical challenges?
 arguably AI poses the most interesting challenges and
questions in computer science today

Today’s Lecture
 What is intelligence? What is artificial intelligence?
 A very brief history of AI
 Modern successes: Stanley the driving robot; Deep Blue the
chess game player; IBM’s Watson computer
 An AI scorecard
 How much progress has been made in different aspects of AI
 AI in practice
 Successful applications
 The rational agent view of AI

What is Intelligence?
 Intelligence:
 “the capacity to learn and solve problems” (Websters
dictionary)
 in particular,
 the ability to solve novel problems
 the ability to act rationally
 the ability to act like humans
 Artificial Intelligence
 build and understand intelligent entities or agents
 2 main approaches: “engineering” versus “cognitive
modeling”

Types of Intelligence
According to Howard Gardner’s multiple intelligence theory,
there are various types of intelligence viz:
 General intelligence: -
 Abilities that allow us to be flexible and adaptive thinkers, not
necessarily tied to acquired knowledge.
 Linguistic-verbal intelligence: -
 Use words and language in various forms / Ability to manipulate
language to express oneself poetically
 Logical-Mathematical intelligence: -
 Ability to detect patterns / Approach problems logically / Reason
deductively
 Musical intelligence: -
 Recognize nonverbal sounds: pitch, rhythm, and tonal patterns
 Spatial intelligence: -
 Typically thinks in images and pictures / Used in both arts and
sciences

Types of Intelligence (2)
 Intrapersonal intelligence: -
 Ability to understand oneself, including feelings and
motivations / Can discipline themselves to accomplish a
wide variety of tasks
 Interpersonal intelligence: -
 Ability to "read people"—discriminate among other
individuals especially their moods, intentions, motivations;
/ Adept at group work, typically assume a leadership role.
 Naturalist intelligence: -
 Ability to recognize and classify living things like plants,
animals
 Bodily-Kinesthetic intelligence: -
 Use one’s mental abilities to coordinate one’s own bodily
movements

Types of Intelligence (3)
Note:
 Understanding the various types of intelligence
provides theoretical foundations for recognizing
different talents and abilities in people
 "What makes life interesting, however, is that we
don’t have the same strength in each intelligence
area, and we don’t have the same amalgam of
intelligences. Just as we look different from one
another and have different kinds of personalities,
we also have different kinds of minds."

Signs of Intelligence
 Learn or understand from experience
 Make sense out of ambiguous or contradictory
messages
 Respond quickly and successfully to new
situations
 Use reasoning to solve problems

More Signs of Intelligence
 Deal with perplexing situations
 Understand and infer in ordinary, rational ways
 Apply knowledge to manipulate the
environment
 Think and reason
 Recognize the relative importance of different
elements in a situation

What is AI? (2)
 There is no agreed definition of the term artificial intelligence.
However, there are various definitions that have been
proposed. These are considered below.
 AI is a study in which computer systems are made that think like
human beings. Haugeland, 1985 & Bellman, 1978.
 AI is a study in which computer systems are made that act like
people.
 AI is the art of creating computers that perform functions that
require intelligence when performed by people. Kurzweil, 1990.
 AI is the study of how to make computers do things, which at the
moment people are better at. Rich & Knight
 AI is the study of computations that make it possible to perceive,
reason and act. Winston, 1992
 AI is considered to be a study that seeks to explain and emulate
intelligent behaviour in terms of computational processes.
Schalkeoff, 1990.
 AI is considered to be a branch of computer science that is
concerned with the automation of intelligent behavior. Luger &
Stubblefield, 1993.

What is AI? (3)
 Artificial Intelligence is the development of
systems that exhibit the characteristics we
associate with intelligence in human behavior:
 perception,
 natural language processing,
 reasoning,
 planning,
 problem solving,
 learning and adaptation,
 etc.

What’s involved in Intelligence?
 Ability to interact with the real world
 to perceive, understand, and act
 e.g., speech recognition and understanding and synthesis
 e.g., image understanding
 e.g., ability to take actions, have an effect
 Reasoning and Planning
 modeling the external world, given input
 solving new problems, planning, and making decisions
 ability to deal with unexpected problems, uncertainties
 Learning and Adaptation
 we are continuously learning and adapting
 our internal models are always being “updated”
 e.g., a baby learning to categorize and recognize animals

Academic Disciplines relevant to
AI
 Philosophy Logic, methods of reasoning, mind as physical
system, foundations of learning, language,
rationality.
 Mathematics Formal representation and proof, algorithms,
computation, (un)decidability, (in)tractability
 Probability/Statistics modeling uncertainty, learning from data
 Economics utility, decision theory, rational economic agents
 Neuroscience neurons as information processing units.
 Psychology/ how do people behave, perceive, process cognitive
Cognitive Science information, represent knowledge.
 Computer building fast computers
engineering
 Control theory design systems that maximize an objective
function over time
 Linguistics knowledge representation, grammars

AI Problems
 What are problems contained in AI?
 Much of the early work focused on formal tasks,
such as game playing and theorem proving.
 E.g. Chess playing, Logic Theorist was an early
attempt to prove mathematical theorems.
 Game playing and theorem proving share the
property that people who do them well are
considered to be displaying intelligence.

AI Problems-Many solution
alternatives
 Despite this it appeared that computers could
perform well at those tasks by being fast at
exploring a large number of solution paths and
then selecting the best one.
 No computer is fast enough to overcome the
combinatorial explosion generated by most
problems.

AI Problems- Based on
commonsense reasoning
 AI focusing on the sort of problem solving we
do every day for instance when we decide to
get to work in the morning, often called
commonsense reasoning.
 In investigating this sort of reasoning Newell,
Shaw, and Simon built the General Problem
Solver (GPS), which they applied to several
commonsense tasks as well performing symbolic
manipulations of logical expression

AI Problems-with domain
knowledge
 However no attempt was made to create a
program with a large amount of knowledge
about a particular problem domain. Only quite
simple tasks were selected.
 As AI research progressed techniques for
handling larger amounts of world knowledge
were developed to deal with problem solving in
specialised domains such as medical
diagnosis and chemical analysis.

AI Problems- perception via
speech, vision, and natural
language
 Perception (vision and speech) is another area
for AI problems.
 Natural language understanding, and problem
solving in specialised domain are other areas
related to AI problems.

AI Problems- Speech based
 The problem of understanding spoken
language is perceptual problem and is hard to
solve from the fact that it is more analog
related than digital related.
 Many people can perform one or may be more
specialised tasks in which carefully acquired
expertise is necessary.

AI Problems-Tasks
 Examples of such tasks include engineering
design, scientific discovery, medical diagnosis,
and financial planning.
 Programs that can solve problems in these
domains also fall under the aegis of Artificial
Intelligence.

AI Problems- List of tasks
 List of the tasks that are targets of works in AI
are:-
 Mundane tasks
 Perception
 Vision
 Speech
 Natural Language
 Understanding
 Generation
 Translation
 Commonsense reasoning
 Robot Control

AI Problems- List of Tasks (2)
 Targets of AI (cont….):
 Formal Tasks
 Games
 Chess, etc.
 Mathematics
 Geometry, Logic, Integral Calculus, etc.
 Expert Tasks
 Engineering
 Design, Fault Finding, Manufacturing Planning
 Scientific analysis
 Medical Diagnosis
 Financial Analysis

AI Problems- Problem-solving
Strategy
 A person who knows how to perform tasks
from several of the categories shown in above
list learn the necessary skills in a standard
order.
 First perceptual, linguistic, and commonsense
skills are learned.
 Later expert skills such as engineering, medicine,
or finance are acquired

AI Problems-Problem-solving
Strategy
 Earlier skills are easier and thus more
amenable to computerised duplication than the
later, more specialised one.
 For this reason much of the initial work in AI
work was concentrated in those early areas.

AI Problems- Success
Scenarios
 The problems areas where now AI is
flourishing most as a practical discipline are
primarily the domains that require only
specialised expertise without the assistance of
commonsense knowledge.
 Expert systems (AI programs) now are up for
day-to-day tasks that aim at solving part, or
perhaps all, of practical, significant problem
that previously required high human expertise.

AI Problems- Fundamental
Questions
 The following four questions need to be
considered before we can progress further:
1) What are the underlying assumptions about
intelligence?
2) What kinds of techniques will be useful for
solving AI problems?
3) At what level if at all can human intelligence be
modelled?
4) When will it be realised when an intelligent
program has been built?

QUESTION#1: The Underlying
Assumption
 A physical symbol system consists of a set of
entities called symbols which are patterns that
can occur as components of another entitiy
called an expression.
 At an instant the system will contain a collection
of these symbol structures. In addition the
system also contains a collection of processes
that operate on expressions to produce other
expressions; processes of creation,
modification, reproduction and destruction.

QUESTION#1: The Underlying
Assumption
 A physical symbol system is a machine that
produces through time an evolving collection
of symbol structures. Such a system is a
machine that produces through time an
evolving collection of symbol structures.
AI system is a physical symbol
system.

Example of Physical Symbol
Systems
 Formal logic: the symbols are words like
"and", "or", "not", "for all x" and so on. The
expressions are statements in formal logic
which can be true or false. The processes
are the rules of logical deduction.
 Algebra: the symbols are "+", "×", "x", "y",
"1", "2", "3", etc. The expressions are
equations. The processes are the rules of
algebra, that allow you to manipulate a
mathematical expression and retain its truth.

Example of Physical Symbol
Systems
 A digital computer: the symbols are zeros and
ones of computer memory, the processes are
the operations of the CPU that change
memory.
 Chess: the symbols are the pieces, the
processes are the legal chess moves, the
expressions are the positions of all the pieces
on the board

Physical Symbol Systems
 The physical symbol system hypothesis
claims that both of these are also examples of
physical symbol systems
 Intelligent human thought: the symbols are
encoded in our brains. The expressions are
thoughts. The processes are the mental
operations of thinking.
 A running artificial intelligence program: The
symbols are data. The expressions are more
data. The processes are programs that
manipulate the data.

The Importance of hypothesis
 The importance of the physical symbol system
hypothesis is twofold.
 It is significant theory about the nature of human
intelligence.
 Forms the basis of the belief that it is possible to
build programs that can perform intelligent tasks
now performed by people.

QUESTION#2: An AI Technique
 Intelligence requires knowledge but knowledge
possesses less desirable properties such as
a) it is voluminous
b) it is difficult to characterize accurately
c) it is constantly changing
d) it differs from data by being organized in a way
that corresponds to its application

QUESTION#2: AI Technique(2)
 An AI technique is a method that exploits
knowledge that is represented so that
 The knowledge captures generalisations ;
situations that share properties, are grouped
together, rather than being allowed separate
representation

 It can be understood by people who must provide
it; although for many programs the bulk of the
data may come automatically, such as from
readings.
 In many AI domains people must supply the
knowledge to programs in a form the people
understand and in a form that is acceptable to the
program.
 It can be easily modified to correct errors and
reflect changes in real conditions

 It can be widely used even if it is incomplete or
inaccurate.
 It can be used to help overcome its own sheer
bulk by helping to narrow the range of possibilities
that must be usually considered.
 In order to characterize an AI technique let us
consider a question answering example.

Example: Question Answering
 Lets discuss a program that reads in English
Text and answers questions, also stated in
English.
 This kind of task is difficult to state formally
and precisely what our problem is and what
constitutes correct solutions to it.

Example: Question
Answering(2)
 For example suppose that the input text were
just the single sentence:
Russia massed troops on the Ukrainian
border
 Then either of the following question-
answering dialogues might occur in the
POLITICS program.

Example: Question
Answering(3)
Dialogue 1
 Q: Why did Russia do this?
 A: Because Russia thought that it could take
political control of Ukraine by sending troops.
 Q: What should the United States do?
 A: The United States should intervene
militarily.

Example: Question
Answering(4)
Dialogue 2
 Q: Why did Russia do this?
 A: Because Russia wanted to increase its
political influence over Ukraine.
 Q: What should the United States do?
 A: The United States should denounce the
Russian action in the United Nations

Example: Question
Answering(5)
 In the POLITICS program answers were
constructed by considering both the input text
and a separate model of the beliefs and
actions of various political entities including
Russia.
 When the model is changed, as it was
between these two dialogues, the system’s
answers also change.

Example: Question
Answering(6)
 In the above example, the first dialogue was
produced when POLITICS was given a model
of conservatives (e.g. views of the Republican
party).
 The second dialogue occurred when
POLITICS program was given a model that
was intended to correspond to the beliefs of a
typical American liberal(e.g. Democratic party).

Example: Question
Answering(7)
 The general point is that defining what it
means to produce a correct answer to a
question may be very hard.
 Usually question-answering form defines what
it means by to be an answer by the procedure
that is used to compute the answer.
 Then their authors appeal to other people to
agree that the answers found by the program
“make sense”.

Conclusion: AI Technique to
Apply
 Three important AI techniques are
 Search – Provides a way of solving problems for
which no more direct approach is available.
 Various solution alternatives have to be explored e.g.
state-space search, adversarial search(aka game
playing)
 Use of Knowledge – Provides a way of solving
complex problems by exploiting the structures of
the objects that are involved.
 Abstraction – Provides a way of separating
important features and variations from the many
unimportant ones that would otherwise
overwhelm any process.

QUESTION#3: The Level of the
Model
 Before starting doing something, it is good
idea to decide exactly what we are trying to do.
We must ask ourselves:
 What is our goal in trying to produce programs
that do the tasks the same way people do?
 Are we trying to produce programs that do the
tasks the same way people do?
 Or are we trying to produce programs that simply
do the tasks in whatever way appears easiest?

Model(2)
 Efforts to build program that perform tasks the
way people do can be divided into two classes
 Those that attempt to solve problems that do
not really fit our definition of AI. i.e. problems
that computer could easily solve.

Model(3)
 The second class attempt to model human
performance are those that do things that fall
more clearly within our definition of AI tasks;
they do things that are not trivial for the computer.
 Reasons for modeling human performance at
these kind of tasks:-

Model(4)
 To test psychological theories of human
performance. E.g. PARRY program written for this
reason, which exploited a model of human paranoid
behaviour to simulate the conversational behaviour
of a paranoid person.
 To enable computer to understand human
reasoning. For example, for a computer to be able
to read a news paper story and then answer
question, such as “Why did Brazil lose the game?”

Model(5)
 To enable people to understand computer
reasoning. In many cases people are reluctant
to rely on the output of computer unless they
can understand how the machine arrived at its
result.
 To exploit what knowledge we can collect from
people.
 To ask for assistance from best performing people
and ask them how to proceed in dealing with their
tasks.

QUESTION#4: Criteria for
success.
 One of the most important questions to answer
in any scientific or engineering research
project is “How will we know if we have
succeeded?”
 So in AI we have to ask ourselves how will we
know if we have constructed a machine that is
intelligent?
 The question is as hard as unanswerable
question “What is Intelligence?”

success(2)
 To measure the progress we use proposed
method known as Turing Test.
 Alan Turing suggested this method to
determine whether the machine can think.
 To conduct this test, we need two people and the
machine to be evaluated. One person act as
interrogator, who is in a separate room from the
computer and the other person.

success(3)
 The interrogator can ask questions of either the
person or computer by typing questions and
receiving typed responses
 However the interrogator knows them only as A
and B and aims to determine which is the person
and which is the machine.
 The goal of the machine is to fool the interrogator
into believing that it is the person. If the machine
succeeds at this, then we will conclude that the
machine can think.

History of AI
 1943: early beginnings
 McCulloch & Pitts: Boolean circuit model of brain
 1950: Turing
 Turing's "Computing Machinery and Intelligence“
 1956: birth of AI
 Dartmouth meeting: "Artificial Intelligence“ name adopted
 1950s: initial promise
 Early AI programs, including
 Samuel's checkers program
 Newell & Simon's Logic Theorist
 1955-65: “great enthusiasm”
 Newell and Simon: GPS, general problem solver
 Gelertner: Geometry Theorem Prover
 McCarthy: invention of LISP

History of AI
 1966—73: Reality dawns
 Realization that many AI problems are intractable
 Limitations of existing neural network methods identified
 Neural network research almost disappears
 1969—85: Adding domain knowledge
 Development of knowledge-based systems
 Success of rule-based expert systems,
 E.g., DENDRAL, MYCIN
 But were brittle and did not scale well in practice
 1986-- Rise of machine learning
 Neural networks return to popularity
 Major advances in machine learning algorithms and applications
 1990-- Role of uncertainty
 Bayesian networks as a knowledge representation framework
 1995-- AI as Science
 Integration of learning, reasoning, knowledge representation
 AI methods used in vision, language, data mining, etc

Success Stories
 Deep Blue defeated the reigning world chess champion Garry
Kasparov in 1997
 AI program proved a mathematical conjecture (Robbins conjecture)
unsolved for decades
 During the 1991 Gulf War, US forces deployed an AI logistics
planning and scheduling program that involved up to 50,000
vehicles, cargo, and people
 NASA's on-board autonomous planning program controlled the
scheduling of operations for a spacecraft
 Proverb solves crossword puzzles better than most humans
 Robot driving: DARPA grand challenge 2003-2007
 2006: face recognition software available in consumer cameras

Example: DARPA Grand
Challenge
 Grand Challenge
 Cash prizes ($1 to $2 million) offered to first robots to complete a long course
completely unassisted
 Stimulates research in vision, robotics, planning, machine learning, reasoning,
etc
 2004 Grand Challenge:
 150 mile route in Nevada desert
 Furthest any robot went was about 7 miles
 … but hardest terrain was at the beginning of the course
 2005 Grand Challenge:
 132 mile race
 Narrow tunnels, winding mountain passes, etc
 Stanford 1st, CMU 2nd, both finished in about 6 hours
 2007 Urban Grand Challenge
 November in Victorville, California

HAL: from the movie 2001
 2001: A Space Odyssey
 classic science fiction movie from 1969
 HAL
 part of the story centers around an intelligent computer called
HAL
 HAL is the “brains” of an intelligent spaceship
 in the movie, HAL can
 speak easily with the crew
 see and understand the emotions of the crew
 navigate the ship automatically
 diagnose on-board problems
 make life-and-death decisions
 display emotions
 In 1969 this was science fiction: is it still science fiction?

Hal and AI
 HAL’s Legacy: 2001’s Computer as Dream and Reality
 MIT Press, 1997, David Stork (ed.)
 discusses
 HAL as an intelligent computer
 are the predictions for HAL realizable with AI today?
 Materials online at
 http://mitpress.mit.edu/e-books/Hal/contents.html
 The website contains
 full text and abstracts of chapters from the book
 links to related material and AI information
 sound and images from the film

Consider what might be involved
in building a computer like Hal….
 What are the components that might be
useful?
 Fast hardware?
 Chess-playing at grandmaster level?
 Speech interaction?
 speech synthesis
 speech recognition
 speech understanding
 Image recognition and understanding ?
 Learning?
 Planning and decision-making?

Fundamentals Challenges-
Where are we?

Can we build hardware as complex as the
brain?
 How complicated is our brain?
 a neuron, or nerve cell, is the basic information processing unit
 estimated to be on the order of 10 12 neurons in a human brain
 many more synapses (10 14) connecting these neurons
 cycle time: 10 -3 seconds (1 millisecond)
 How complex can we make computers?
 108 or more transistors per CPU
 supercomputer: hundreds of CPUs, 1012 bits of RAM
 cycle times: order of 10 - 9 seconds
 Conclusion
 YES: in the near future we can have computers with as many basic
processing elements as our brain, but with
 far fewer interconnections (wires or synapses) than the brain
 much faster updates than the brain
 but building hardware is very different from making a computer behave like a
brain!

Can Computers Talk?
 This is known as “speech synthesis”
 translate text to phonetic form
 e.g., “fictitious” -> fik-tish-es
 use pronunciation rules to map phonemes to actual sound
 e.g., “tish” -> sequence of basic audio sounds
 Difficulties
 sounds made by this “lookup” approach sound unnatural
 sounds are not independent
 e.g., “act” and “action”
 modern systems (e.g., at AT&T) can handle this pretty well
 a harder problem is emphasis, emotion, etc
 humans understand what they are saying
 machines don’t: so they sound unnatural
 Conclusion:
 NO, for complete sentences

Can Computers Recognize
Speech?
 Speech Recognition:
 mapping sounds from a microphone into a list of
words
 classic problem in AI, very difficult
 “Lets talk about how to wreck a nice beach”
 (I really said “________________________”)
 Recognizing single words from a small
vocabulary
 systems can do this with high accuracy (order of 99%)
 e.g., directory inquiries
 limited vocabulary (area codes, city names)
 computer tries to recognize you first, if unsuccessful hands
you over to a human operator
 saves millions of dollars a year for the phone companies

Recognizing human speech
(ctd.)
 Recognizing normal speech is much more difficult
 speech is continuous: where are the boundaries between words?
 e.g., “John’s car has a flat tire”
 large vocabularies
 can be many thousands of possible words
 we can use context to help figure out what someone said
 e.g., hypothesize and test
 try telling a waiter in a restaurant:
“I would like some dream and sugar in my coffee”
 background noise, other speakers, accents, colds, etc
 on normal speech, modern systems are only about 60-70%
accurate
 Conclusion:
 NO, normal speech is too complex to accurately recognize
 YES, for restricted problems (small vocabulary, single speaker)

Can Computers Understand
speech?
 Understanding is different to recognition:
 “Time flies like an arrow”
 assume the computer can recognize all the words
 how many different interpretations are there?

speech?
 1. time passes quickly like an arrow?
 2. command: time the flies the way an arrow times the flies
 3. command: only time those flies which are like an arrow
 4. “time-flies” are fond of arrows

speech?
 1. time passes quickly like an arrow?
 2. command: time the flies the way an arrow times the flies
 3. command: only time those flies which are like an arrow
 4. “time-flies” are fond of arrows
 only 1. makes any sense,
 but how could a computer figure this out?
 clearly humans use a lot of implicit commonsense knowledge in
communication
 Conclusion: NO, much of what we say is beyond the
capabilities of a computer to understand at present

Can Computers “see”?
 Recognition v. Understanding (like Speech)
 Recognition and Understanding of Objects in a scene
 look around this room
 you can effortlessly recognize objects
 human brain can map 2d visual image to 3d “map”
 Why is visual recognition a hard problem?
 Conclusion:
 mostly NO: computers can only “see” certain types of objects under limited
circumstances
 YES for certain constrained problems (e.g., face recognition)

Can Computers Learn and
Adapt ?
 Learning and Adaptation
 consider a computer learning to drive on the freeway
 we could teach it lots of rules about what to do
 or we could let it drive and steer it back on course when it heads for
the embankment
 systems like this are under development (e.g., Daimler Benz)
 e.g., RALPH at CMU
 in mid 90’s it drove 98% of the way from Pittsburgh to San Diego without any
human assistance
 machine learning allows computers to learn to do things without
explicit programming
 many successful applications:
 requires some “set-up”: does not mean your PC can learn to forecast the
stock market or become a brain surgeon
 Conclusion: YES, computers can learn and adapt, when
presented with information in the appropriate way

Can computers plan and make optimal
decisions?
 Intelligence
 involves solving problems and making decisions and plans
 e.g., you want to take a holiday in Brazil
 you need to decide on dates, flights
 you need to get to the airport, etc
 involves a sequence of decisions, plans, and actions
 What makes planning hard?
 the world is not predictable:
 your flight is canceled or there’s a backup on the 405
 there are a potentially huge number of details
 do you consider all flights? all dates?
 no: commonsense constrains your solutions
 AI systems are only successful in constrained planning problems
 Conclusion: NO, real-world planning and decision-making is still beyond the
capabilities of modern computers
 exception: very well-defined, constrained problems

Summary of State of AI Systems in
Practice
 Speech synthesis, recognition and understanding
 very useful for limited vocabulary applications
 unconstrained speech understanding is still too hard
 Computer vision
 works for constrained problems (hand-written zip-codes)
 understanding real-world, natural scenes is still too hard
 Learning
 adaptive systems are used in many applications: have their limits
 Planning and Reasoning
 only works for constrained problems: e.g., chess
 real-world is too complex for general systems
 Overall:
 many components of intelligent systems are “doable”
 there are many interesting research problems remaining

Intelligent Systems in Your
Everyday Life
 Post Office
 automatic address recognition and sorting of mail
 Banks
 automatic check readers, signature verification systems
 automated loan application classification
 Customer Service
 automatic voice recognition
 The Web
 Identifying your age, gender, location, from your Web surfing
 Automated fraud detection
 Digital Cameras
 Automated face detection and focusing
 Computer Games
 Intelligent characters/agents

AI Applications: Machine
Translation
 Language problems in international business
 e.g., at a meeting of Japanese, Korean, Vietnamese and Swedish investors, no common
language
 or: you are shipping your software manuals to 127 countries
 solution; hire translators to translate
 would be much cheaper if a machine could do this
 How hard is automated translation
 very difficult! e.g., English to Russian
 “The spirit is willing but the flesh is weak” (English)
 “the vodka is good but the meat is rotten” (Russian)
 not only must the words be translated, but their meaning also!
 is this problem “AI-complete”?
 Nonetheless....
 commercial systems can do a lot of the work very well (e.g.,restricted vocabularies in
software documentation)
 algorithms which combine dictionaries, grammar models, etc.
 Recent progress using “black-box” machine learning techniques

What’s involved in Intelligence?
(again)
 Perceiving, recognizing, understanding the real
world
 Reasoning and planning about the external world
 Learning and adaptation
 So what general principles should we use to
achieve these goals?

Characteristics of AI
 Symbolic Processing
 AI emphasizes manipulation of symbols rather than
numbers.
 The manner in which symbols are processed is non-
algorithmic since most human reasoning process do
not necessarily follow a step by step approach
(algorithmic approach).
 Heuristics
 Are similar to rules of thumb where you need not
rethink completely what to do every time a similar
problem is encountered.
 Inferencing
 This is a form of reasoning with facts and rules using
heuristics or some search strategies.

Characteristics of AI (2)
 Pattern matching
 A process of describing objects, events or
processes in terms of their qualitative features and
logical and computational relationships.
 Knowledge Processing
 Knowledge consists of facts, concepts, theories,
heuristics methods, procedures and relationships.
 Knowledge bases.
 Collection of knowledge related to a problem or an
opportunity used in problem.
 Reasoning occurs based on this knowledge base.

Contrasting AI with Natural
Intelligence
 Important commercial advantages of AI are:-
1) AI is permanent as long as computer system and
programs remain unchanged
2) AI offers ease of duplications and dissemination as
compared to long apprenticeship for natural
intelligence.
3) AI can be less expensive than natural intelligence.
4) AI being a computer system is consistent and
thorough; natural intelligence may be erratic since
people are erratic, they don’t perform consistently.
5) AI can execute certain tasks much faster than
humans can.
6) AI can perform certain tasks better than many or
even most people.

Contrasting AI with Natural
Intelligence (2)
Natural Intelligence has the following
advantages
1) Natural intelligence is creative, AI is
uninspired- human always determine
knowledge.
2) Natural intelligence enables people to
benefit from use of sensory experience
directly, while most AI systems must work
with symbolic knowledge.

Different Types of Artificial
Intelligence
1. Modeling exactly how humans actually think
2. Modeling exactly how humans actually act
3. Modeling how ideal agents “should think”
4. Modeling how ideal agents “should act”
 Modern AI focuses on the last definition
 we will also focus on this “engineering” approach
 success is judged by how well the agent performs

Thinking humanly
 Cognitive Science approach
 Try to get “inside” our minds
 E.g., conduct experiments with people to try to “reverse-
engineer” how we reason, learning, remember, predict
 Problems
 Humans don’t behave rationally
 e.g., insurance
 The reverse engineering is very hard to do
 The brain’s hardware is very different to a computer
program

Thinking rationally
 Represent facts about the world via logic
 Use logical inference as a basis for reasoning about these
facts
 Can be a very useful approach to AI
 E.g., theorem-provers
 Limitations
 Does not account for an agent’s uncertainty about the world
 E.g., difficult to couple to vision or speech systems
 Has no way to represent goals, costs, etc (important aspects of
real-world environments)

Acting rationally
 Decision theory/Economics
 Set of future states of the world
 Set of possible actions an agent can take
 Utility = gain to an agent for each action/state pair
 An agent acts rationally if it selects the action that
maximizes its “utility”
 Or expected utility if there is uncertainty
 Emphasis is on autonomous agents that behave
rationally (make the best predictions, take the best
actions)
 on average over time
 within computational limitations (“bounded rationality”)

Symbolic AI
 Based on Newell & Simons Physical Symbol System Hypothesis
 Uses logical operations that are applied to declarative knowledge
bases (FOPL)
 Commonly referred to as “Classical AI”
 Represents knowledge about a problem as a set of declarative
sentences in FOPL
 Then logical reasoning methods are used to deduce consequences
 Another name for this type of approach is called “the knowledge-
based” approach
 The Symbol Processing Approach uses “top-down” design of
intelligent behavior.

Sub-symbolic Approach
 Based on the Physical Grounding Hypothesis
 “bottom-up” style
 Starting at the lowest layers and working upward.
 In the sub-symbolic approach signals are generally used rather
than symbols
 Proponents believe that the development of machine intelligence
must follow many of the same evolutionary steps.
 Sub-symbolic approaches rely primarily on interaction between
machine and environment. This interaction produces and emergent
behavior (evolutionary robotics, Nordin, Lund)
 Some other sub-symbolic approaches are: Evolutionary
Computation, Artificial Immune Systems, and Neural Networks

Turing Test for Intelligence
 Tests the ability of a computer system to act humanly
 The aim is to determine if the human interrogator
thinks he/she is communicating with a human.
 Turing (1950) "Computing machinery and
intelligence“
 "Can machines think?"  "Can machines behave
intelligently?“
 To pass Turing Test the computer must:
 Process natural language;
 Represent knowledge;
 Reason;
 Learn and adapt to the new situations.
 Total Turing test included vision & robotics.

Turing Test (the Imitation game)-
1950

Modelling an AI System
 A typical AI system consists of three
subsystems, i.e.,
 Perception Subsystem
 Reasoning Subsystem
 Action Subsystem(made of actuators/effectors)

AI application areas
 Game Playing
 Much of the early research in state space search was
done using common board games such as checkers,
chess, and the 15-puzzle
 Games can generate extremely large search spaces.
These are large and complex enough to require
powerful techniques for determining what alternative
to explore

 Automated reasoning and Theorem Proving
 Theorem-proving is one of the most fruitful branches
of the field
 Theorem-proving research was responsible in
formalizing search algorithms and developing formal
representation languages such as predicate calculus
and the logic programming language

 Expert System
 One major insight gained from early work in problem
solving was the importance of domain-specific
knowledge
 Expert knowledge is a combination of a theoretical
understanding of the problem and a collection of
heuristic problem-solving rules
 Current deficiencies:
 Lack of flexibility; if human cannot answer a question
immediately, he can return to an examination of first
principle and come up with something
 Inability to provide deep explanations
 Little learning from experience

 Natural Language Understanding and
Semantics
 One of the long-standing goals of AI is the
creation of programming that are capable of
understanding and generating human language

 Modeling Human Performance
 Capture the human mind (knowledge
representation)

 Robotics
 A robot that blindly performs a sequence of
actions without responding to changes or being
able to detect and correct errors could hardly be
considered intelligent
 It should have some degree of sensors and
algorithms to guide it

 Machine Learning
 Learning has remained a challenging area in AI
 An expert system may perform extensive and
costly computation to solve a problem; unlike
human, it usually doesn’t remember the solution
 Examples include:
 Decision tree learning
 Genetic algorithms
 Neural networks

Application Domains of AI
 Application domain areas include:
 Military
 Medicine
 Industry
 Entertainment
 Education
 Business

Summary of Today’s Lecture
 Intelligence and types
 Artificial Intelligence involves the study of:
 automated recognition and understanding of signals
 reasoning, planning, and decision-making
 learning and adaptation
 AI has made substantial progress in
 recognition and learning
 some planning and reasoning problems
 …but many open research problems
 AI Applications
 improvements in hardware and algorithms => AI applications in industry, finance, medicine,
and science.
 Rational agent view of AI
 Two views of AI: symbol based vs. sub-symbolic AI
Turing test for intelligence and AI applications

Lecture1-fundamentals.pptx

Recommended

Recommended

More Related Content

Similar to Lecture1-fundamentals.pptx

Similar to Lecture1-fundamentals.pptx (20)

Recently uploaded

Recently uploaded (20)

Lecture1-fundamentals.pptx