Transcript What is AI…? - Department of Computing

What is AI…?
Dr. Simon Colton
Computational Bioinformatics Laboratory
Department of Computing
Imperial College, London
What isn’t Artificial Intelligence
AI in PopSci books and the Media

Kevin “March of the Machines” Warwick
– Robots will take over the Earth

Roger “Emperors New Mind” Penrose
– Computers will never be intelligent
What isn’t AI

Mark “The Human Computer” Jeffery
– Computers will evolve to be human
?

Ray “The Age of Spiritual Machines” Kurzweil
– Humans will evolve to be computers
Two Restricted Views of AI

As a tool to study (human) intelligence
– Just the latest part of the philosophers toolbox
– Mostly scientific

As a set of methods for solving problems
– Which take intelligence to solve in humans
– Mostly technological

In reality, AI encompasses both of these
– Part science, part technology
Two Characterisations of AI
“What problem
do I have?”
“How on earth can I get
my machine to do
clever things?”
Characterisation by Problem

If you know the type of problem
– There are established techniques to use

Some problems you may want solving:
– Translating, proving, learning, optimising, …
– Seeing, hearing, speaking, moving, …

This is how AI is usually taught
– And how subjects are arranged in textbooks
Considerations for Problem Solving

How to specify the problem
– So the computer knows when it’s done

How to represent solutions
– Representation, representation, representation
– Symbolic and non-symbolic

How to search for solutions
– Calculation, simple search, rules of thumb
An Example: Computer Maths

Would you do this by hand if you had a
calculator: 171717 * 98765?
– If we can get a computer to do it,
• Then it’s extremely reliable

Computers do more complicated maths:
e.g., 17 < x < 19, 15 < x+y < 20, 13 < y-x < 17

And can beat humans sometimes:
– I wanted to prove, that, in ring theory:
• (all x, (x+x = x*x))  (all w x (((w*w)*x)*(w*w)) = id)
– I couldn’t prove this, but Otter could!
A Nicer Characterisation

As answers to:
– “How can I get my machine to be clever”

Seven answers over the years:
–
–
–
–
–
–
–
Use logic
Use introspection
Use brains
Use evolution
Use the physical world
Use society
Use ridiculously fast computers
Elementary, my dear Watson

Logical approach
– Idea: represent and reason

“It’s how we wish we solved
problems…
– Just like Sherlock”

Very well respected
– Established
• 3000 years of development
– Techniques for reasoning
• Deduction & induction
– Programming languages
Introspection

Logic has limits
– Combinatorial explosion

“Maybe we’re not logical
– But we are intelligent”

Use introspection
– Can be highly effective
– Can be problematic

Heuristic search
– Using rules of thumb to
guide the solving process
BrainWare

“Maybe we don’t know our
psychology
– But it’s our brains which do
the intelligent stuff”

And we do know
– Some neuroscience

Idea is to build:
– Artificial Neural Networks
– Simulate neurons firing
• Networks configuring themselves

Mostly used for prediction
– E.g., stock markets (badly)
Evolve or Perish

“Our brains give us our smarts,
– But what gave us our brains?”

Idea: evolve programs
– Simulate reproduction and
survival of fittest

Problem Solving:
– Genetic algorithms (parameters)
– Genetic programming (program)

Artificial Life
– Can we evolve “living” things
The More the Merrier

“We live and work in societies
– Each of us has a job to do”

Idea to simulate society
– Autonomous agents

Each has a subtask
– Together solve the problem

Agencies have structure
 Agents can
– compete, co-operate, haggle,
argue, …
The Harsh Realities of Life
“But we evolved intelligence
for a reason”
 Idea: get robots to do simple
things in the physical world

– Dynamic & dangerous

From survival abilities
– Intelligence will evolve

Standing up is much more
intelligent than
– Translating French to German
– In Evolutionary terms
Brute Force

“Let’s stop being so
clever and use
computers to their full”
– Processor/memory gains
have been enormous

Can solve problems in
“stupid” ways
– Relying on brute force

The Deep Blue way
– Little harsh on IBM
An Example: RHINO

Robotic museum tour guide
– Robot + computers
– And worried researchers
• Who didn’t intervene

Highly successful
– 18.6 kilometres, 47 hours
– 50% attendance rise
– 1 tiny mistake
• No breakage/injury

Great science
– Using many approaches
– Won best paper award
Where is AI?
In industry – see Rob’s talk
 In education – see Andrew’s talk
 In research

– Computing, psychology, philosophy,
– Cognitive science, linguistics, biology,
– Mathematics, physics, …

Artificial Intelligence does not class
itself as simply a subset of computing
Some Aspirations

“Big” AI
– Building of human-level intelligence
into robots like Lieutenant Commander Data

“Small” AI
– Get computers to undertake some intelligent tasks
– Mostly problem solving
– But sometime more creative “artefact generation”
• Painting pictures, composing melodies, writing poems, …
– This is what most of us do
Computers can Create

Produced by the NeVar system © Machedo

Uses Genetic Programming
 Evolve the program to draw these
 Evolutionary
Art is very big
Resources…

This is meant to stimulate questions