Download CS 8520: Artificial Intelligence

CS 8520: Artificial Intelligence
Introduction
Paula Matuszek
Fall, 2008
CSC 8520 Fall, 2008. Paula Matuszek
1
AI Course Details
• Instructor:
– Paula Matuszek
• [email protected]
• 610-270-6851
• Course web page
–
–
–
–
There will be one. Still working on where, exactly.
Syllabus, Requirements
Handing in Homework
Academic Integrity
• Required and recommended texts
• Questions?
• Student questionnaire
CSC 8520 Fall, 2008. Paula Matuszek
2
Our Approach
• Following the book, mostly:
– Tools and techniques (through chapter 10)
– Some of the domains, depending on interest
• Working in the lab:
– We will spend some part of most classes doing handson stuff. Trying out tools and applications, exploring
what's out there, etc.
• AI is also FUN, exciting, always new. I hope to
convey some of why.
• We will all get more out of this class if you speak
up. I encourage questions and ideas and
discussion in class.
CSC 8520 Fall, 2008. Paula Matuszek
3
Class Background
• In order to help structure and focus the
course, we need to have an idea of the
interests and backgrounds of the members
of the class.
–
–
–
–
Name
Something about your background
Something about why you're interested in AI
Something about what you hope to get from
this class
CSC 8520 Fall, 2008. Paula Matuszek
4
Resources
• We will add to the class web page lists of
interesting resources. Two major sources
you should be aware of:
– Our textbook is in extensive use, and there is a
web page with many resources and links at
aima.cs.berkeley.edu
– The American Association for Artificial
Intelligence is the primary professional
organization in the US for AI. Their web page
at www.aaai.org has many resources.
CSC 8520 Fall, 2008. Paula Matuszek
5
What is AI?
• There are no crisp definitions
Q. What is artificial intelligence?
A. It 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.
Q. Yes, but what is intelligence?
A. 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.
(John McCarthy, 1956. http://www.formal.Stanford.EDU/jmc/whatisai )
CSC 8520 Fall, 2008. Paula Matuszek
6
Other possible AI definitions
• AI is a collection of hard problems which can be solved
by humans and other living things, but for which we
don’t have good algorithmic solutions
– e.g., understanding spoken natural language, medical
diagnosis, circuit design, etc.
• AI Problem + Sound theory = Engineering problem
• Many problems used to be thought of as AI but are now
considered not
– e.g., compiling Fortran in 1955, symbolic mathematics in
1965, image cleanup, Optical character recognition.
CSC 8520 Fall, 2008. Paula Matuszek
??
7
Ways to Examine the field of AI
• The field of can generally be viewed from two
directions:
– The techniques you use
•
•
•
•
Search
Knowledge Representation
Inference
Logic
– The areas you're working in
•
•
•
•
•
Planning
Learning
Natural Language Understanding
Games
Etc. Etc. Etc.
CSC 8520 Fall, 2008. Paula Matuszek
8
What’s easy and what’s hard?
• Easier: many of the high level tasks we usually associate
with “intelligence” in people
– e.g., Symbolic integration, proving theorems, playing
chess, medical diagnosis, etc.
• Harder: tasks that lots of animals can do
–
–
–
–
–
walking around without running into things
catching prey and avoiding predators
interpreting complex sensory information
modeling the internal states of other animals from their behavior
working as a team (e.g. with pack animals)
• What's the difference?
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
9
History
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
10
Current State
• Is AI a failure? Is AI dead?
• NO. AI is
– pervasive
– invisible
• AI researchers sometimes joke that there are
no solved problems in AI. Why? Once
they're solved they aren't AI any more.
CSC 8520 Fall, 2008. Paula Matuszek
11
Foundations of AI
Mathematics
Economics
Psychology
CSC 8520 Fall, 2008. Paula Matuszek
Computer
Science &
Engineering
AI
Cognitive
Science
Philosophy
Biology
Linguistics
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
12
Why AI?
• Engineering: To get machines to do a wider variety
of useful things
– e.g., understand spoken natural language, recognize
individual people in visual scenes, find the best travel plan
for your vacation, etc.
• Cognitive Science: As a way to understand how
natural minds and mental phenomena work
– e.g., visual perception, memory, learning, language, etc.
• Philosophy: As a way to explore some basic and
interesting (and important) philosophical questions
– e.g., the mind body problem, what is consciousness, etc.
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
13
Possible Approaches
Like
humans
Think
GPS
Act
Eliza
CSC 8520 Fall, 2008. Paula Matuszek
Well
Rational
agents
AI tends to
work mostly
in this area
Heuristic
systems
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
14
Think well
Like
humans
• Develop formal models of
knowledge representation,
reasoning, learning,
memory, problem solving, that
can be rendered in algorithms.
• There is often an emphasis on
systems that are provably
correct, and guarantee finding
an optimal solution.
CSC 8520 Fall, 2008. Paula Matuszek
Think
Act
Well
GPS
Rational
agents
Eliza
Heuristic
systems
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
15
Act well
Like
humans
Think
GPS
• For a given set of inputs, generate an
appropriate output that is not necessarily
correct but gets the job done.
Eliza
Act
• A heuristic (heuristic rule, heuristic
• method) is a rule of thumb, strategy, trick,
• simplification, or any other kind of device
• which drastically limits search for solutions
• in large problem spaces.
• Heuristics do not guarantee optimal solutions; in fact, they
do not guarantee any solution at all: all that can be said
for a useful heuristic is that it offers solutions which are
good enough most of the time.
– Feigenbaum and Feldman, 1963, p. 6
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
Well
Rational
agents
Heuristic
systems
16
Like
humans
Think like humans
Think
GPS
Well
Rational
agents
• Cognitive science approach
• Focus not just on behavior and I/O Act Eliza Heuristic
systems
but also look at reasoning process.
• Computational model should reflect "how" results
were obtained.
• Provide a new language for expressing cognitive
theories and new mechanisms for evaluating them
• GPS (General Problem Solver): Goal not just to
produce humanlike behavior (like ELIZA), but to
produce a sequence of steps of the reasoning
process that was similar to the steps followed by a
person in solving the same task.
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
17
Act like humans
Like
humans
Think
•
•
•
•
•
Behaviorist approach.
Not interested in how you get
results, just the similarity to what
human results are.
Exemplified by the Turing Test
(Alan Turing, 1950).
CSC 8520 Fall, 2008. Paula Matuszek
Act
Well
GPS
Rational
agents
Eliza
Heuristic
systems
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
18
Turing Test
• Three rooms contain a person, a computer, and an
interrogator
• The interrogator can communicate with the other
two by teleprinter.
• The interrogator tries to determine which is the
person and which is the machine.
• The machine tries to fool the interrogator into
believing that it is the person.
• If the machine succeeds, then we conclude that the
machine can think.
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
19
Some Turing Test - like Programs
• Over the course of the last 60 years there
have been a number of programs which
might be said to have at least tried the
Turing Test, although not always with that
goal.
CSC 8520 Fall, 2008. Paula Matuszek
20
Eliza
• ELIZA: A program that simulated a psychotherapist
interacting with a patient and successfully "passed" the
Turing Test.
• Coded at MIT during 1964-1966 by Joel Weizenbaum.
• First script was DOCTOR.
– The script was a simple collection of syntactic patterns not unlike
regular expressions
– Each pattern had associated reply including bits of the input (after
simple transformations (my -> your)
• Weizenbaum was shocked at reactions:
– Psychiatrists thought it had potential.
– People unequivocally anthropomorphised.
– Many thought it solved the NL problem.
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
21
Eliza
• This is a transcript of a conversation with an Eliza-like program found in Emacs.
Try typing Meta-x doctor in your favorite Emacs or Googling for Eliza.
•
•
•
•
•
•
•
•
•
•
•
•
•
I am the psychotherapist. Please, describe your problems.
I am angry.
Are you angry often?
No.
Why not?
Usually I am even-tempered.
Why do you say usually you are even tempered?
I do not get angry usually.
What causes you to be angry?
My mother.
Tell me something about your family.
My mother is a busybody.
You seem to dwell on your family.
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
22
Colby’s PARRY
• Kenneth Colby modeled
a paranoid using the
same techniques circa
1968.
• PARRY has basic
emotions. If it gets
angry, its replies
become more hostile.
• In a Turing test, expert
psychiatrists were
unable to distinguish
PARRY’s ramblings
from those of real
paranoids.
CSC 8520 Fall, 2008. Paula Matuszek
H: Good afternoon. My name is
Doctor Jones.
P: Pleased to meet you.
H: How are you?
P: All right I guess.
H: Could you tell me your name?
P: Pat Smith.
H: I though you might have some
problems.
P: I've been upset lately.
H: By what?
P: People get on my nerves
sometimes.
H: Tell me more.
P: Do you know anything about
bookies?
...
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
23
The Loebner Contest
• A modern version of the Turing Test, held annually,
with a $100,000 cash prize.
• http://www.loebner.net/Prizef/loebner-prize.html
• Restricted topic (removed in 1995) and limited time.
• Participants include a set of humans and a set of
computers and a set of judges.
• Scoring
– Rank from least human to most human.
– Highest median rank wins $2000. ($3000 in 2005)
– If better than a human, win $100,000. (Nobody yet…)
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
24
So when WILL we decide that
computers are intelligent?
CSC 8520 Fall, 2008. Paula Matuszek
25
How Do We Know When We're There?
• Some requirements I think any test we use must
meet:
– Whatever test we use must not exclude the majority of
adult humans. I can't play chess at a grand master
level!
– Whatever test we use must produce an observable or
testable result. "Isn't intelligent because it doesn't have
a mind" is perhaps a topic for interesting philosophical
debate, but it's not of any practical help.
• AI from a computer scientist perspective! Not the
Chinese Room
CSC 8520 Fall, 2008. Paula Matuszek
26
What can AI systems do
In the meantime, AI can be an effective tool. Here are some
example applications of current AI capabilities:
• Computer vision: face recognition from a large set
• Robotics: autonomous (mostly) car
• Natural language processing: simple machine translation
• Expert systems: medical diagnosis in a narrow domain
• Spoken language systems: ~1000 word continuous speech
• Planning and scheduling: Hubble Telescope experiments
• Learning: text categorization into ~1000 topics
• User modeling: Bayesian reasoning in Windows help
• Games: Grand Master level in chess (world champion),
checkers, etc. Some skill in Go.
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
27
What can’t AI systems do yet?
• Understand natural language robustly (e.g., read
and understand articles in a newspaper)
• Surf the web
• Interpret an arbitrary visual scene
• Learn a natural language
• Construct plans in dynamic real-time domains
• Refocus attention in complex environments
• Perform life-long learning
CSC 8520 Fall, 2008. Paula Matuszek
Based on http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
28
What's Happening Now in AI?
• Homework assignment will explore some of the
things now going on in AI
• A useful resource in current AI news is
http://www.aaai.org/AITopics/newstopics/main.html
Note: Many of the slides of this presentation are modified from those of
Professor Marie DesJardins, University of Maryland Baltimore County.
The originals can be found at
http://www.cs.umbc.edu/courses/graduate/671/fall05/slides/c1_intro.ppt
CSC 8520 Fall, 2008. Paula Matuszek
29
First Homework Assignment
1.
3.
•
From the textbook: Answer questions 1.2, 1.11 and any 5 of 1.7.
Look at the other questions and think about them; you might find it
interesting to make note of your thoughts and read them again at the
end of the course. For question 1.2, you can find a copy of Turing's
paper at http://www.abelard.org/turpap/turpap.htm.
Skim through your textbook, including the detailed contents list.
Choose two chapters from chapters 11-27 that you are most
interested in seeing us cover in class.
Academic Integrity revisited.
CSC 8520 Fall, 2008. Paula Matuszek
30