Artificial Intelligence

(AI)
CS-401

Dr. Hafeez Ur Rehman

(Email: [email protected])
1
 Text Book
• Text book:
 Artificial Intelligence: A Modern Approach
by Stuart Russell & Peter Norvig
• Recommended books:
 Artificial Intelligence and Intelligent
Systems, N.P. PADHY
 Practical Artificial Intelligence Programming
with java, Mark Watson

2
 Course Outline
Goal is to introduce you to set of Artificial
Intelligence (AI) concepts & techniques. The
course will cover broader AI concepts like:
 Agents & Environments
 Search Strategies
 Logic and Knowledge Representation
 Planning
 Reasoning under Uncertainty
 Machine Learning
3
 First Week
 Introduction
– What is AI? (and why is it so cool?)
– Brief history of AI
– What’s the state of AI now?
– Examples
– Challenges (What AI can & can’t do!)

4
 What is AI??
Artificial Intelligence

5
 What is Intelligence?
• Intelligence:
– “The capacity to learn and solve problems”
(Websters dictionary)
– in particular,
• the ability to solve novel problems
• the ability to think & act like humans
• the ability to think & act rationally

• Artificial Intelligence
– build and understand intelligent entities or agents
– Two main approaches: “engineering” versus
“cognitive modeling” 6
 What’s involved in Intelligence?
1. 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
2. Reasoning and Planning
– modeling the external world, given input
– solving new problems, planning, and making decisions
– ability to deal with unexpected problems, uncertainties
3. 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
7
 Artificial Intelligence as Science
 Physics: Where did the physical universe
come from and what laws guide its
dynamics?
 Biology: How did biological life evolve and
how do living organisms function?
 Artificial Intelligence: What is the nature
of “intelligence” and what constitutes
intelligent behavior?
8
 Artificial Intelligence as Engineering
 How can we make software and machines more
powerful, adaptive, and easier to use?
Examples:
Speech recognition
Natural language understanding
Computer vision and image understanding
Intelligent user interfaces
Data mining
Mobile robots, Softbots, Humanoids
Medical expert systems…

9
 What is AI? A simple Definition
 AI is the attempt for the reproduction of human
reasoning and intelligent behavior by
computational methods
Reasoning is the process of
thinking about something in
Intelligent a logical way by establishing
and verifying facts to draw
behavior
inferences or conclusions.
Computer
Intelligent behavior is how
a person performs " in
response to questions and
problems the answers to
Humans which are NOT immediately
known" - Arthur L. Costa
10
 What is AI?

Four possible ways to define Artificial

Intelligence i.e.

A discipline that systematizes and automates

reasoning processes to create machines that:

Act like humans Act rationally

Think like humans Think rationally

11
 Act like humans Act rationally
Think like humans Think rationally
 The goal of AI is to create computer systems that
perform tasks regarded as requiring intelligence when
done by humans.
 In this AI Methodology we take a task at which people are better,
e.g.:
• Prove a theorem
• Play chess
• Plan a surgical operation
• Diagnose a disease
• Navigate in a building
and build a computer system that does it automatically
 But HUMANS MAKE ERRORS,,,,,do we want to duplicate or copy
human imperfections? 12
 Machine Test: Act Like Humans
• The Turing Test

• What capabilities would a computer need to have to pass

the Turing Test?
• Natural language processing
• Knowledge representation
• Automated reasoning
• Machine learning
• Turing predicted that by the year 2000, machines would
be able to fool 30% of human judges for five minutes
13
A. Turing, Computing machinery and intelligence, Mind 59, pp. 433-460, 1950
Example Eliza – A partial success…
• 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 an associated reply which might include bits of
the input (after simple transformations (my  your)
• Weizenbaum was shocked at reactions:
– Psychiatrists thought it had potential.
– People unequivocally related it to humans like response.
– Many thought it solved the NL problem.
14
Another Application of the Turing Test
 CAPTCHA: Completely Automatic Public
Turing tests to tell Computers and
Humans Apart
 E.g.:
• Display visually distorted words
• Ask user to recognize these words
 Example of application: Have only
humans do the activity e.g., open email
account.
15
 AI is solved? Eugene Goostman

http://www.bbc.com/news/technology-27762088
http://en.wikipedia.org/wiki/Eugene_Goostman
http://www.scottaaronson.com/blog/?p=1858 17
 What’s wrong with the Turing test?
• Variability in protocols, judges
• Success depends on deception!
• Chatbots can do well using “cheap tricks”
• First example: ELIZA (1966)
• Chinese room argument: one may simulate
intelligence without having true intelligence
(more of a philosophical objection)

18
A better Turing test?

19
http://www.newyorker.com/online/blogs/elements/2013/08/why-cant-my-computer-understand-me.html
A better Turing test?
• Multiple choice questions that can be
easily answered by people but cannot be
answered by computers using “cheap
tricks”:

The trophy would not fit in the brown suitcase

because it was so small.

What was so small?

• The trophy
• The brown suitcase
H. Levesque, On our best behaviour, IJCAI 2013
Published in International Joint Conference on Artificial Intelligence 20
A better Turing test?
• Multiple choice questions that can be easily
answered by people but cannot be answered
by computers using “cheap tricks”:

The trophy would not fit in the brown suitcase

because it was so large.

What was so large?

• The trophy
• The brown suitcase

21
H. Levesque, On our best behaviour, IJCAI 2013
A better Turing test?
• Advantages over standard Turing test
• Test can be administered and graded by machine
• Does not depend on human subjectivity
• Does not require ability to generate English sentences
• Questions cannot be evaded using verbal dodges
• Questions can be made “Google-proof” (at least for now…)

22
H. Levesque, On our best behaviour, IJCAI 2013
 Act like humans Act rationally
Think like humans Think rationally

 In this type of approach to AI, how the computer performs

tasks does matter and are inspired by human thinking.
 The reasoning steps are important…… done in the same
fashion as humans process thoughts…
 Ability to create and manipulate symbolic knowledge
(definitions, concepts, theorems, …)
 Example GPS (General Problem Solver): solved simple
problems such as the Towers of Hanoi that could be
sufficiently formalized. Problem: Explosion of states.
 What is the impact of hardware on low-level reasoning, e.g.,
to go from signals to symbols?
23
 Act like humans Act rationally
Think like humans Think rationally

• Cognitive Science approach

– Try to get “inside” our minds
– e.g., conduct experiments with people to try to
“reverse-engineer” how we reason, learn,
remember, predict
• Problems
– Humans don’t behave rationally
• e.g., insurance, politically motivated debate
– The reverse engineering is very hard to do
– The brain’s hardware is very different to a
computer program
24
AI definition 02: Think Like Humans
• Need to study the brain as an information
processing machine: cognitive science and
neuroscience

25
AI definition 02: Think Like Humans
• Can we build a brain?

26
Source: L. Zettlemoyer
AI definition 02: Think Like Humans

27
 Act like humans Act rationally
&
Think like humans Think rationally

 In this (rational) type of approach to AI, the goal is to

build agents that always make the “best” decision
given what is available (knowledge, time, resources)
 “Best” means maximizing the expected value of a
utility function
 What is the impact of self-consciousness, emotions,
desires, love for music, fear of dying, etc ... on human
intelligence? Are they part of Human Intelligence?

28
AI definition 03: Thinking rationally
• Idealized or “right” way of thinking
• Logic: patterns of argument that always yield correct
conclusions when supplied with correct premises
• “Socrates is a man; all men are mortal; therefore Socrates is
mortal.”
• Logicist approach to AI: describe problem in formal
logical notation and apply general deduction
procedures/rules to solve it
• Problems with the logicist approach
1. Computational complexity of finding the solution
2. Describing real-world problems and knowledge in logical
notation; becoz not all facts are 100% true.
3. Dealing with uncertainty/expected value
4. A lot of “rational” behavior has nothing to do with logic
29
AI definition 04: Acting rationally
• A rational agent acts to optimally achieve its
goals
• Goals are application-dependent and are expressed in
terms of the utility of outcomes
• Being rational means maximizing your (expected)
utility
• This definition of rationality only concerns the
decisions/actions that are made, not the
cognitive process behind them
• In practice, a rational agent acts to achieve best
outcome or when there is uncertainty the best
expected outcome.
30
 History

Ada Lovelace-1842
31
 History of AI
• 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
32
 History of AI
• 1966—73: Reality dawns / WINTER OF AI
– Realization that many AI problems are intractable (difficult to handle or
manage e.g. Speech Recognition, NLP etc.)
• 1969—85: Adding domain knowledge
– Development of knowledge-based systems
– Success of rule-based expert systems,
• E.g., DENDRAL (mol. Structure predictor), MYCIN (blood infections)
• But were brittle (less adaptable) 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 is opted as a Science
– Integration of learning, reasoning, knowledge representation, vision,
NLP etc. 33
 After All this Debate the question remains….
Can Machines Act/Think Intelligently?

 Yes, if intelligence is narrowly defined as

information processing
AI has made impressive achievements showing that
tasks initially assumed to require intelligence can be
automated
 Maybe yes and maybe not, if intelligence is not
separated from the rest of “being human”

34
Success Stories AI Systems

35
 Self-driving cars

http://www.technologyreview.com/featuredstory/520431/d Google plans to assemble a test fleet of

riverless-cars-are-further-away-than-you-think/ 100 electric two-seat prototypes in 2015

Uber employee of the future:

The self-driving car
36
 Speech and natural language

• Instant translation with Word Lens

• Have a conversation with Google Translate

37
http://www.skype.com/en/translator-preview/ http://googleblog.blogspot.com/2015/01/hallo-hola-ola-
more-powerful-translate.html
 Vision

• Computer Eyesight Gets a Lot More Accurate,

NY Times Bits blog, August 18, 2014
• Building A Deeper Understanding of Images,
Google Research Blog, September 5, 2014
• Baidu caught gaming recent supercomputer
performance test, Engadget, June 3, 2015
38
Games
• 1997: IBM’s Deep Blue defeats the reigning world
chess champion Garry Kasparov
• 1996: Kasparov Beats Deep Blue
“I could feel – I could smell – a new kind
of intelligence across the table.”
• 1997: Deep Blue Beats Kasparov
“Deep Blue hasn't proven anything.”
• 2007: Checkers is solved
• Though checkers programs had been beating the best human
players for at least a decade before then
• 2014: Heads-up limit Texas Hold-em poker is solved
• First game of imperfect information
• 2015: Deep reinforcement learning beats humans at
classic arcade games
39
Mathematics
• In 1996, a computer program written by researchers
at Argonne National Laboratory proved a
mathematical conjecture unsolved for decades
• NY Times story: “[The proof] would have been called
creative if a human had thought of it”
• Mathematical software:

40
Logistics, scheduling, planning

• 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 Remote Agent software operated the
Deep Space 1 spacecraft during two
experiments in May 1999
• In 2004, NASA introduced the MAPGEN
system to plan the daily operations for the
Mars Exploration Rovers

41
Robotics
• Mars rovers
• Autonomous vehicles
• DARPA Grand Challenge
• Self-driving cars
• Autonomous helicopters
• Robot soccer
• RoboCup
• Personal robotics
• Humanoid robots
• Robotic pets
• Personal assistants?

42
DARPA Robotics Challenge (2015)

http://www.popularmechanics.com/technology/robots/a1590 43
7/best-falls-from-darpa-robot-challenge/
 Towel-folding robot

YouTube Video

• J. Maitin-Shepard, M. Cusumano-Towner, J. Lei and P. Abbeel, Cloth

Grasp Point Detection based on Multiple-View Geometric Cues with
Application to Robotic Towel Folding, ICRA 2010 44
• More clothes folding
 What AI systems can’t 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
 Play Go well Done in 2016.
 Construct plans in dynamic real-time domains
 Refocus attention in complex environments
 Perform life-long learning

45
The End

46