Introduction

Artificial Intelligence
 Artificial Intelligence

Intelligence e
i v
…What is er ce
P
it???
n d i nk
er sta T h
Un d
A
Lea dapt to ne
Rem rn f w situ
em b rom ations
eAI
r is one of the most exciting exareas
peri of
Computer Science enc
Solve u e
nseen p on
roblems ea s
to r
t y
bi li
A
Can machines
think???
 Humans Vs. Machines
Humans Machines
• Symbolic calculation • Numeric calculation
• Natural language • Machine instructions
understanding (0s and 1s)
• Not very precise • Precise
• Knowledge • Data
• Generalize from • Cannot generalize
examples from examples
• Deal with noisy inputs • Cannot deal with noise

There are now systems that can achieve the human attributes to an extent
 Computer Vision

Machine learning

Game playing Information retreival

Text processing

ARTIFICIAL INTELLIGENCE

Speech processing

Symbolic AI Logic and reasoning

 Birth of AI
After the invention of the number crunching
machines scientists turned their attention to building
machines that have some form of intelligence
Informally research in AI started centuries ago when philosophers and
thinkers ventured on understanding the human mind

After world war II many scientists started work on making intelligent

machines. Alan Turing was one of the first

The word ‘AI’ was coined by John McCarthy in Dartmouth in 1956

AI is not exactly coming from ONLY computer science….it has origins in

math, logic, philosophy, psychology, biology, cognitive science etc. etc.

AI has produced important results in Computer

Science, Psychology and many other disciplines
 What is AI???
•AI wants to duplicate human characteristics
like creativity, language use, self-improvement
and learning,
•AI has become a branch of computer science
But takes its inspiration from many other areas
like biology, math, psychology, etc., etc.
•AI wants to build machines that will function
independently in a complex, changing world

Agent : Something capable of perceiving its environment and taking actions that
maximize its chances of success. They exhibit intelligent behavior e.g. robots
(physical agents) or software programs (virtual agents)
 TWO MORE THINGS…

• Hard AI/Strong AI
A machine should
exhibit intelligence
same as human

• Soft AI/WEAK AI
A machine is intelligent
in a specific domain.
Soft AI adds more
functionality to a
computer and a useful
tool for humans…(not
so ambitious as hard AI)
 Examples of an AI System
Eliza : Hello, I am Eliza.
You : Hello i am Kamran Eliza : It emulates a
Eliza : How long have you been kamran? computer therapist
You : 33 years
Eliza : Tell me more...
You : what do you want to know? Try talking to Eliza
Eliza : I'm not sure I understand you fully. yourself
You : You said “tell me more”. I am saying what is it you
want to know?
Eliza: name? http://
You: Kamran www.manifestation.com
Eliza : Names don't interest me.
You : Really what interests you?
/neurotoys/eliza.php3
Eliza : We were discussing you, not me.
You : why...i want to discuss you
Eliza : You're not really talking about me, are you?
You : yes
Eliza : Are you Sure?
You : yes
Eliza : Please don't repeat yourself!
You : ok you are boring me now
Eliza : What makes you think I am boring you now?
You : because i know it
Eliza : Say, do you have any psychological problems?
 Monty Hall Problem
• Suppose you're on a game show, and you're given the
choice of three doors: Behind one door is a car; behind
the others, goats. You pick a door, say No. 1, and the
host, who knows what's behind the doors, opens another
door, say No. 3, which has a goat. He then says to you,
"Do you want to pick door No. 2?" Is it to your
advantage to switch your choice?

Can AI program with

knowledge of probability
theory help you decide
better????

Source: Wikipedia
 GAME PLAYING….CHESS
MECHANICAL TURK (1770-1854)

December 2006
• Deep Fritz, a chess-playing computer, has
beaten human counterpart world chess
champion Vladimir Kramnik in a six-
game battle in Bonn, Germany.
http://news.bbc.co.uk/2/hi/europe/6212076.stm and wikepe
dia.com
 SPEECH RECOGNITION
• First system was Audrey by Bell labs in 1952 for
digits only
• 1962-Shoebox by IBM, understood 16 words
• 1970s…CMU Harpy understood 1011 words,
which is approximately the vocabulary of a 3
year old
• 1990s…Dragon and “Dragon naturally
speaking” for continuous speech. Required 45
minutes of training
 SIRI
• Siri on iPhone
lets you use
your voice to
send messages,
schedule
meetings, place
phone calls, and
more.

www.apple.com
 RECENT ADVANCES…
ALVINN
• Computer vision system
• By CMU Lab
• Computer controlled minivan and used to
navigate across US. It was in control of steering
the vehicle 98% times. Humans took over 2%
times, mostly at exit ramps.
AUTONOMOUS PLANNING & SCHEDULING
NASA’s remote agent program became the first
on-board autonomous planning program to
control and schedule operations of a spacecraft.

Slide 14
Mehreen Saeed
 DRIVERLESS CARS
• Google has acquired the
license for driverless cars
in Nevada (May 2012)
• Test fleet of around 8
vehicles
– Will have a red number
plate and an infinity
symbol

www.forbes.com
Slide 15
Mehreen Saeed
DARPA GRAND CHALLENGE
• Challenge by defense “advanced research
projects agency” for driverless cars
– DARPA 2007: Urban challenge where vehicle
was to be driven automatically in a city like
environment avoiding all obstacles and
interacting with the regular traffic and
obeying all traffic rules
Winner was collaborative effort
between CMU and general motors
Google Crunches One Trillion Pieces of
Data With Single Click
• Powerful data analysis tool called Dremel
• Analyzes information on huge amounts of
information at an amazing speed

www.techinvestornews.com
 APPROACHES TO AI

Systems that THINK Systems that THINK

LIKE HUMANS RATIONALLY
Cognitive modeling Laws of thought
approach approach

Systems that ACT LIKE Systems that ACT

HUMANS RATIONALLY
Turing test approach Rational agent approach

Resource: Russell and Norvig

 HISTORY OF AI
1941 First electronic computer
1943 McCulloch & Pitts artificial neuron
1951 First neural network (Minsky & Edmonds)
1956 Birth of AI. Coined by McCarthy Dartmouth conference

1956 Logic theorist

1958 Lisp
1965 Eliza
1967 Dendrall (chemical analysis system)
1971 SHRDLU (robot arm carried out instructions in English. Moved around blocks)

1972 Prolog
1974 MYCIN (Formally termed first expert system). Showed power of rule based systems

1980 Blackboard model for speech understanding

Mid 80 Backprop for neural nets
90s Data mining, scheduling, virtual reality and AI information extraction systems
1997 Deep Blue AI system beats human chess master
 2000 & Beyond…
• Robopets (smart toys)…sociable machines that also
express emotions
• Open-Cyc project: AI project to assemble comprehensive
database and ontology of everyday common sense
knowledge
– Has over a million human defined rules, assertions and common
sense ideas
– Based on predicate calculus
– Syntax similar to Lisp
(Ontology: Description of concepts and relations that can
exist for an agent or a number of agents…describe
individuals, classes, relations and attributes)
• Many advances in speech recognition
• Many advances in computer vision
• Many advances in natural language processing
=