AI - A Gentle Introduction

ARTIFICIAL INTELLIGENCE
A Gentle, Jargon-Free Introduction
K. Raghunathan
Chartered Engineer
FIETE, MIE, MCSI, MISTE, MISTD, MIIMM
Retd Dy Controller (R&D) [MoD]
Section 1
What is AI ?
• Study/Science of
– Intelligent Machines
– Making Machines Intelligent
AI & Computers
• Computers seem to have some Intelligence
– Solve Math problems [Arithmetic, Algebra, Calculus,
Geometry etc]
– Do text processing
– Control Eqpt & Processes in Industry
– Communicate
• Natural Belief : computers can be made more
intelligent
• So, AI is considered a Branch of Computer
Science
Machines & Humans
• Right from Cavemen, humans are
inventing new Tools, Instruments &
Machines
– To do all our hard work
– Even for entertainment
– To reduce/avoid our physical/mental strain
Human Tasks
• What are our everyday tasks ?
– Professional
– Personal
Functions of our Sensory Organs
• Eye
– See
• Ear
– Hear
• Tongue / Mouth
– Taste, Speak
• Nose
– Smell, Breathe
• Skin / Limbs
– Feel, do Physical Work – Walk, Climb, Lift, Throw, Eat etc
Sixth Sense
• Comprehend [Understand]
– Vision
– Language
• Remember
• Plan
• Decide
• Learn
• Emote
Brain & Mind
• Concrete & Abstract
• Tangible & Intangible
• Similar to Computer
• Like Hardware & Software
What we still do ourselves ?
• Logical Thinking & Reasoning
• Making wise decisions
– often with Incomplete, Vague & Uncertain
facts at hand
AI attempts to fill this gap

Intelligent Machines
• Machines that can
– Think
– Decide
– Solve
– Plan
– Learn
• Thinking Machines
• Artificial Intelligence
Can Machines Think ?
• Some believe so & some don’t
• What Level of Intelligence will make a

Machine qualify as Intelligent Machine or
Thinking Machine ?
• The Turing Test

The Turing Test
The Turing Test
• A machine can be deemed intelligent
• If it can make a human think that
• he/she is talking to a human
• & not to a machine
Intelligent Machines
• A machine can be considered Intelligent

• if it can solve complex problems
• requiring a high level of intelligence
• normally found only in human beings
AI – Science or Art ?
• Science
– Multi-disciplinary & Inter-disciplinary
• Art
– Very Intricate Fine Art
– Like Painting & Sculpting
• Science
– involves designing machines
• Art
– involves programming
Inputs to AI
• Maths [Symbolic Logic]
• Physiology [Working of Body Parts of Living
Organisms]
• Philosophy & Psychology [Human Logic &
Behaviour]
• Cognitive Science [Perception, Understanding]
• Cybernetics [Communication & Control in
Animals & Machines]
• Electronics [Making Machines]
• Computer Science [Programming the Machine's
Behaviour]
AI’s Output
• Used by almost all Sciences

– Including Computer Science
• Maximum dependency on AI
– Robotics & Neural Networks
– Considered part of AI
Section 2
Back to Square One
What is Intelligence ?
Let us have your Opinions please

Intelligence
• Abstract
• Eludes accurate, comprehensive definition
• Easier to understood than to define
• Seems to be directly proportional to

Knowledge
Intelligence (cont’d)
• The Albert Einstein Anecdote

Intelligence (contd)
• Knowledge [Lots of it !]
• Ability to put knowledge to use
• Pertains to what we can do with our 5

Senses
• Pertains more to what we can do with our
6th Sense !
What is Knowledge ?
Anybody wish to define ?
Knowledge
• Abstract
• Eludes accurate, comprehensive definition
• Easier to understand than to define
• AI has to deal with various aspects of

Knowledge
Knowledge (contd)
• Classification
• Representation [including Storage in
Computer Memory]
• Acquisition [Learning]
• Manipulation
– Searching & Matching
– Reasoning [Inference, Drawing Conclusions,
Decision Making, Problem Solving]
• Planning
Classification of Knowledge
• Factual or Procedural
• Certain or Uncertain
• Consistent or Inconsistent [vary with time,
place, person or situation]
• Well-Defined [Crisp] or Vague [Fuzzy]
• Complete or Incomplete [Partial]
Reasoning Techniques
• Depend on the Type of Knowledge being
dealt with
• Certain, consistent & crisp : Predicate
Logic
• Uncertain : Probablistic Reasoning
• Inconsistent : Truth Maintenance Systems
• Fuzzy : Fuzzy Logic
• Partial : Statistical & Probablistic
Reasoning
Reasoning Techniques (cont’d)
• Ad hoc Methods
– Closed World Assumption

– Circumscription
– Abductive Reasoning
– Modal & Temporal Logics
Learning Methods
• By Rote [by Reading]

• By being Taught
• From Experience
• By Analogy [From Examples]
• By Drawing Conclusions
– From other known Facts
What Humans are Good at
An Interactive Experiment
2
What Humans are Good at
• Humans seem to be good at

– Reasoning with Uncertain, Fuzzy & Partial
Knowledge
– Learning from Experience & Examples
• AI Research pays more attention to these
Areas
AI Techniques
• Procedures & Algorithms employed in AI
• Compared to those used in other Sciences
– Not merely Formulae
– Use different approaches
– More efficient, effective & intelligent
– Profuse use of Heuristics
• Heuristic
– “Educated Guess”
– Born out of Past Experience
Heuristics
- An Interactive Experiment
[opening a book at a desired page]
Heuristic Search Techniques
• Many Methods available
• Most Noteworthy are:

– A*
– MiniMax with Alpha-Beta Pruning
Expert Systems
&
Decision Support Systems
• A Speciality of AI
• Very High Level of Intelligence & Domain
Knowledge
• a Level normally found only in Experts in
that Field
Expert Systems
Search
Engine Query
User
Knowledge Match
Inter
Logic
face
Base
Inference Response
Logic
Section 3
Miscellaneous Topics
• Miscellaneous, but nevertheless Essential
• Automation
– Useful in Hazardous Environs
• Computer Vision
– Object Recognition
• Natural Language Processing
• Pattern Recognition & Pattern Classification
– Data Mining, Image Processing
• Guide/Help/Rescue
• Weather/Business Forecasting
• Neural Networks
Vision
• Object Recognition
• Range Estimation
• Emotion Recgnition
Natural Language Processing
• Vocabulary
• Ambiguity
– Human Languages are Context-Sensitive
– One word having many meanings
– Many words having same/similar meaning
– Accent varies from place to place
• Real Life Examples
• Feelings/Emotions
AI Milestones
• Too many to list here
• Early days of AI
– Mycin - Medical Diagnosis
– R - Design of Computer HW Architecture
– Deep Blue - Chess Champion
– Eliza - Conversationist
– Racter - Writes Books
AI Milestones (2)
• Modern days:
• Expert Systems & Decision Support Systems in

a wide range of fields
– especially Medicine, Finance & Entertainment
• Wide range of Medical Eqpt - CAT Scan etc
• Analysis & Prediction - Economics, Stockmarket,
Agriculture, Genetics, Elections
• 3D Computer Games & Walk-throughs
• Helpful Robots - Rescue, Operating in
Hazardous environments
AI Milestones (3)
• Spell-Check & Word Completion
facility in Text Processing Programs
• Voice Commands to open/close files etc
• Programs to read text, to teach Languages
• 3D & Walk-Through Games/Presentations
Basic Hurdle
• Intelligent Machine = Thinking Machine
• To make a machine think, act & react like a

human:
• We should first understand how humans think,
act & react
• Why a person thinks, acts & reacts in a particular
way
• Why different persons think, act & react
differently
• Why same person thinks, acts & reacts differently
at different times, even if situations are similar
Basic Hurdle (cont’d)
• Real life examples

– Bernard Shaw & Isadora Duncan Episode
– Once I took my wife to a movie ….
• Unfortunate Reality for AI scientists
• God Almighty has a habit of making
humans
– so varied,
– so un-understandable
– so unpredictable !
• Given such a starting point
– How are we going to make machines that
would behave like humans ?
• Everyone says “AI is a difficult subject".
– Not only Students, even Faculty say so
– except the Enthusiastic & Optimistic AI
Researcher, of course
• The Ground Reality is:

– Humans are so difficult to understand !
• WE are our own Hurdles !

Concluding Remarks
• In spite of the Enormity of the Obstacles, AI
marches ahead, relentlessly
• The progress is slow
• New Findings & Achievements are Sporadic, few
& far between
• But that is no reason to lose hope
– We have achieved many things, which were once
considered impossible
– We have set foot on the Moon
• Let us hope that one day we will see
The Ultimate Intelligent Machine
Any Questions ?
Thank You !

AI - A Gentle Introduction

Uploaded by

Copyright:

Available Formats

AI - A Gentle Introduction

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

AI - A Gentle Introduction

Uploaded by

Copyright:

Available Formats

ARTIFICIAL INTELLIGENCE

A Gentle, Jargon-Free Introduction

AI attempts to fill this gap

• What Level of Intelligence will make a

• The Turing Test

• A machine can be considered Intelligent

• Used by almost all Sciences

Let us have your Opinions please

• Seems to be directly proportional to

• The Albert Einstein Anecdote

• Pertains to what we can do with our 5

• AI has to deal with various aspects of

– Closed World Assumption

• By Rote [by Reading]

• Humans seem to be good at

• Most Noteworthy are:

• Expert Systems & Decision Support Systems in

• To make a machine think, act & react like a

• Real life examples

• The Ground Reality is:

• WE are our own Hurdles !

You might also like