Artificial Intelligence
1.
Introduction
a.
Artificial
Intelligence (coined in 1956, John
McCarthy)
i.
apply to the use
of computers for studying and modeling problem-solving skills once thought only
to used by humans
ii.
play games, proof
theorems, translated natural language, learned from their experiences
b.
Turing’s 1950
paper “Computing Machinery and Intelligence”
i.
first suggest
computers be used to simulate human intelligence
ii.
Is there that
much similarities between the brain and a computer
1.
storage – memory
2.
ability to follow
steps
3.
input/output
4.
process sensory
perceptions
5.
etc.
6.
7.
Can computers
think?
c.
HAL 9000 “Hal”
i.
skills and
attributes that allow him to conform to any def of AI
2.
Intelligent Automata
a.
automaton (def.) – used to describe anything that acts on its
own. (Gr. lit. self-moving)
i.
Pre 20th century
– simply mimicked humans motions and actions
1.
clay figures
2.
great clocks –
with moving figures
3.
The Turk – a
chess playing machine 19th cent.
ii.
20th
century – (electronics and circuits) – thinking machine?
1.
still only small
advances – more of an electronic mimicking
iii.
1950 – the
computer
1.
Alan Turing –
“Can machines think?”
a.
a test: the imitation game (Turing Test)
i.
2 humans and a
computer
ii.
interrogator and responser (human and computer)
iii.
computer is said
to have superior intelligence if the interrogator is fooled
iv.
If the computer
acts intelligently then it is intelligent
iv.
Modern
1.
no consensus as
to what is AI
2.
Definitions:
a.
Minsky – “AI is the science of making machines do things
that would require intelligence if done by man.”
i.
Turing – it does
make a difference how a machine is intelligent
ii.
focus on
algorithms and programming techniques
b.
Hayes – “the
study of intelligence as computation”
i.
Extent to which a
humans can be considered “computers”
1.
they are
interested in how humans solve the problems and use sensory input
c.
Tessler – “whatever hasn’t been done yet”
i.
emphasizes the elusive natures of both intelligence and
computation.
ii.
as soon as it can be computerized then “boom” – not
really intelligent.
iii.
True AI - moving
target
3.
People and Machines (What AI is not)
a.
Thinking
Effortlessly
i.
not simply
cognition (cognition def. the mental process of knowing – such as
awareness, perception, reasoning, judgment)
1.
hardiest things
to simulate are those things we do effortlessly
a.
Easy for us/Hard
for computer:
i.
walk, speak,
vision recognition
b.
Hard for us/Easy
for computer:
i.
find a long list of prime numbers, pattern recognition,
code breaker, etc.
c.
Our natural
ability to deal with vision and natural language and etc – “we are wired to
adept at these tasks”
d.
One area of
research is giving computer the facilities to learn and grow.
e.
God moment (we
are created in the image of God) – not evolution moment
2.
walking, talking,
pattern recognition, sound processing, speaking
b.
Thinking Deeply
i.
analogy and
metaphor
1.
we live our lives
based on metaphors – this situation or thought or new thing is like
_______________
2.
“a human being, even a child, ‘knows’ vastly more than any
computer yet built.”
3.
Examples of
analogy:
a.
“Time change and
going to church” story
b.
learning a new
game
i.
What would it
like to have to re-teach a child the basics of game-playing (carrying over some of the basics from one
context to the next)
1.
“There is a goal”
2.
“There is an
opposing team”
3.
“There are rules”
ii.
We make
appropriate assumptions all the time – because we realize when something is
“like” something else.
iii.
Is this why young
children are constantly asking the simplest questions – they are forming their
analogy bank
c.
Thinking Hard
i.
when faced with a situation with no (or little) analogy –
we must reason.
ii.
relying on a large stock of acquired knowledge.
iii.
progress in this
area (reason)
1.
computers are
quintessential (def. the essence of a thing in its purest) logic machines
2.
playing in a
computer’s home court – logical reason
a.
rules of logic
are simple to program
i.
special languages
PROLOG (also LISP)
1.
designed around
logic
2.
designed
especially for AI
3.
sample program -
change([H,Q,D,N,P]) :-
member(H,[0,1,2]), /* Half-dollars */ member(Q,[0,1,2,3,4]), /* quarters */ member(D,[0,1,2,3,4,5,6,7,8,9,10]) , /* d */ member(N,[0,1,2,3,4,5,6,7,8,9,10, /* ni*/ 11,12,13,14,15,16,17,18,19,20]), S is 50*H + 25*Q +10*D + 5*N, S =< 100, P is 100-S.
3.
chess, checkers, backgammon,
Othello
4.
scheduling, medical diagnostic knowledge , etc.
d.
Thinking about
Computers
i.
drawbacks: lack
of knowledge about human intellect
ii.
instead of trying
to produce intelligent machine – understand intelligence
iii.
use computers to
understand intelligence (experimental cognition)
iv.
the computers
becomes the test bed
1.
you can’t dissect
the human brain
e.
Comparison:
brain and computer
i.
storage – brain
(50 trillion) vs. computer (1 trillion)
ii.
complexity
1.
parallel
processing
a.
brain – each
neuron connected to 5,000 others
i.
millions of
processors – each connected to 1000’s
ii.
transfer of data
– slower due to chemical transfer 1000 ft/s
b.
computer
i.
1000’s of
processors each connected to 100’s
ii.
transfer data –
million times faster
iii.
cycle time –
(switch to change) million x’s faster
iv.
Blue Gene
iii.
speed
a.
brain
i.
transfer of data
– slower due to chemical transfer 1000 ft/s
b.
computer
i.
transfer data –
million times faster
ii.
cycle time –
(switch to change) million x’s faster
iv.
conclusion
1.
computers are
much faster for doing – simple, repetitive, serial tasks (crunching raw data)
2.
humans are faster doing – complex, high level, and parallel
tasks.
4.
Artificial Skills
a.
Intro
b.
Language
Processing
i.
Language –
distinguishes humans from other species
1.
fruitful insight
into human intellect
2.
since 1950s –
translating one language into another
a.
solution to
“worldwide language problem
b.
large bilingual
dictionaries
c.
more difficult
then first thought
i.
English to
Russian
1.
English “the
spirit is willing is willing, but the flesh is weak”
2.
Russian “the
vodka is acceptable, but the meat has spoiled”
3.
Shifted toward
language understanding
a.
working knowledge
of elements of the language + grammatical structure
i.
– contributed to
the development of high-level languages
b.
incorporates
i.
parsing
1.
different parts
and grammatical correct
ii.
sense of word
based on surrounding words
iii.
an extensive and shared
knowledge about the real world – transcends language
iv.
Difficult – “Time
flies like an arrow; fruit flies like a banana”
4.
Chatterbots - http://www.botspot.com/search/s-chat.htm
a.
ELIZA – using scripted
dialogue
i.
re-phrasing to
get question and using neutral responses
c.
Speech
Recognition
i.
isolated word
detectors
ii.
Specific context
speech recognition - http://shop.voicerecognition.com/items.asp?Cc=DRAGON&source=google
iii.
Many difficulties
both technical and theoretical
d.
Knowledge
processing
i.
Early
game-playing and problem solving
1.
state-space
description
e.
Visual processing
i.
image enhancement
ii.
edge detection
iii.
OCR (Optical
Character Recognition)
f.
Learning
i.
neural networks
ii.
training