84
Artificial Artificial Intelligence Intelligence and and Computing Science Computing Science October 19, 2012

Artificial Intelligence and Computing Science

  • Upload
    leola

  • View
    30

  • Download
    0

Embed Size (px)

DESCRIPTION

Artificial Intelligence and Computing Science. October 19, 2012. AI in the Seventies and Eighties. A belated realization: in most situations, intelligent behaviour requires knowledge. How many things do you know?. ``I can ’ t think of a hundred things that I know a thousand things about ’’ - PowerPoint PPT Presentation

Citation preview

Page 1: Artificial  Intelligence and  Computing Science

Artificial IntelligenceArtificial Intelligenceand and

Computing ScienceComputing Science

October 19, 2012

Page 2: Artificial  Intelligence and  Computing Science

AI in the Seventies and Eighties

A belated realization: in most situations, intelligent behaviour requires knowledge

Page 3: Artificial  Intelligence and  Computing Science

How many things do you know?

``I can’t think of a hundred things that I know a thousand things about’’

(Marvin Minsky, `Semantic Information Processing’, 1968)

Page 4: Artificial  Intelligence and  Computing Science

AI in the Seventies

Winograd, SHRDLU

Expert Systems

Minsky, `Frames’

Rumelhart and McClellan, `Parallel Distributed Programming’ (return of the Perceptron)

Page 5: Artificial  Intelligence and  Computing Science
Page 6: Artificial  Intelligence and  Computing Science

SHRDLU’s World

Page 7: Artificial  Intelligence and  Computing Science

Sample Dialogue

Winograd: What’s on the green block?

Shrdlu: A pyramid

Winograd: Put the red block on the green block

Shrdlu: OK

Winograd: Why did you move the purple pyramid?

Shrdlu: To clear off the green block

Winograd: Why did you put the red block on the green block?

Shrdlu: You told me to

Page 8: Artificial  Intelligence and  Computing Science

Winograd: Pick up a big red block

Shrdlu: OK

Winograd: Grasp the pyramid

Shrdlu: I don’t understand which pyramid you mean

Winograd: Find a block which is taller than the one you are holding and put it into the box

Shrdlu: By `it’, I assume you mean the block which is taller than the one I am holding.

OK.

Page 9: Artificial  Intelligence and  Computing Science

Winograd: What does the box contain?

Shrdlu: The blue pyramid and the blue block

Winograd: What is the pyramid supported by?

Shrdlu: The box

Winograd: Can the table pick up blocks?

Shrdlu: No

Page 10: Artificial  Intelligence and  Computing Science

Things SHRDLU doesn’t know:

`red’ and `green’ are colours

SHRDLU’s world is flat

A table has legs but no arms

SHRDLU is a computer program

Page 11: Artificial  Intelligence and  Computing Science

Expert Systems (rule-based)

Knowledge can be represented by a number of `if…then’ rules plus an

inference engine.

E.g, ``IF temperature is high AND rash is present, THEN patient has measles.’’

Page 12: Artificial  Intelligence and  Computing Science

We can extract the rules from human experts via interviews.This process is known as `knowledge engineering’:

Page 13: Artificial  Intelligence and  Computing Science

`If an animal has fur, it is a mammal’`If an animal has feathers, it is a bird’`If an animal is a bird, it can fly’`If an animal has scales, it is a fish’`If an animal is a fish, it can swim’`If an animal lays eggs and has fur, it is a duck-billed platypus’

This gives us a set of rules that an inference engine (or `expert system shell’) can reason about.

Two popular modes of reasoning are forward chaining and backward chaining:

Page 14: Artificial  Intelligence and  Computing Science

`If an animal has fur, it is a mammal’`If an animal has feathers, it is a bird’`If an animal is a bird, it can fly’`If an animal is a bird, it lays eggs’`If an animal has scales, it is a fish’`If an animal is a fish, it can swim’`If an animal lays eggs and has fur, it is a duck-billed platypus’

Forward chaining:

Given a new fact (`Tweety has feathers’), search for all matching conditionals,draw all possible conclusions, and add them to the knowledge base:

:- Tweety is a bird:- Tweety can fly:- Tweety lays eggs

Potential problem: we run into the combinatorial explosion again

Page 15: Artificial  Intelligence and  Computing Science

`If an animal has fur, it is a mammal’`If an animal has feathers, it is a bird’`If an animal is a bird, it can fly’`If an animal is a bird, it lays eggs’`If an animal has scales, it is a fish’`If an animal is a fish, it can swim’`If an animal lays eggs and has fur, it is a duck-billed platypus’`Tweety has feathers’

Backward chaining:

Given a query (`Does Tweety lay eggs?’),search for all matching consequentsand see if the database satisfies the conditionals:

Page 16: Artificial  Intelligence and  Computing Science

`If an animal has fur, it is a mammal’`If an animal has feathers, it is a bird’`If an animal is a bird, it can fly’`If an animal is a bird, it lays eggs’`If an animal has scales, it is a fish’`If an animal is a fish, it can swim’`If an animal lays eggs and has fur, it is a duck-billed platypus’`Tweety has feathers’

Backward chaining:

`Does Tweety lay eggs?’

Page 17: Artificial  Intelligence and  Computing Science

`If an animal has fur, it is a mammal’`If an animal has feathers, it is a bird’`If an animal is a bird, it can fly’`If an animal is a bird, it lays eggs’`If an animal has scales, it is a fish’`If an animal is a fish, it can swim’`If an animal lays eggs and has fur, it is a duck-billed platypus’`Tweety has feathers’

Backward chaining:

`Does Tweety lay eggs?’`Is Tweety a bird?’

Page 18: Artificial  Intelligence and  Computing Science

`If an animal has fur, it is a mammal’`If an animal has feathers, it is a bird’`If an animal is a bird, it can fly’`If an animal is a bird, it lays eggs’`If an animal has scales, it is a fish’`If an animal is a fish, it can swim’`If an animal lays eggs and has fur, it is a duck-billed platypus’`Tweety has feathers’

Backward chaining:

`Does Tweety lay eggs?’`Is Tweety a bird?’Does Tweety have feathers?’

Page 19: Artificial  Intelligence and  Computing Science

Backward chaining:

This method is used by Prolog, for example

Conclusion: Yes, Tweety does lay eggs

Page 20: Artificial  Intelligence and  Computing Science

`If an animal has feathers, it is a bird’`If an animal is a bird, it can fly’`If an animal is a bird, it lays eggs’

Potential problem: A lot of rules have exceptions.

Page 21: Artificial  Intelligence and  Computing Science

Frames

(Marvin Minsky, 1974)

Page 22: Artificial  Intelligence and  Computing Science

A frame allows us to fill in defaultknowledge about a situation from apartial description. For example,

``Sam was hungry. He went into aMcdonalds and ordered a hamburger.Later he went to a movie.’’

Did Sam eat the hamburger?

Page 23: Artificial  Intelligence and  Computing Science
Page 24: Artificial  Intelligence and  Computing Science

So we can economically represent knowledge by defining properties at the most general level, then letting specific cases inherit those properties…

Event

Transaction

Buying something

Buying a hamburger

Page 25: Artificial  Intelligence and  Computing Science

Return of the perceptron(now called a `neural net’)

Changes since 1969:

Hidden layers

Non-linear activation function

Back-propagation allows learning

Page 26: Artificial  Intelligence and  Computing Science

Rumelhart and McClelland`Parallel Distributed Processing’

Use neural nets to represent knowledge by the strengths of associations between different concepts, rather than as lists of facts, yielding programs that can learn from example.

Page 27: Artificial  Intelligence and  Computing Science

Conventional Computer Memory

Register One 01100110

Register Two 11100110

Register Three 00101101

. . . .

Page 28: Artificial  Intelligence and  Computing Science
Page 29: Artificial  Intelligence and  Computing Science

AI: 1979-2000

Douglas Lenat, `CYC’,

Douglas Hofstadter, `Fluid Analogies’

Brian Hayes, `Naïve Physics’

Page 30: Artificial  Intelligence and  Computing Science
Page 31: Artificial  Intelligence and  Computing Science
Page 32: Artificial  Intelligence and  Computing Science

CYC’s data are written in CycL, which is adescendant of Frege’s predicate calculus(via Lisp).

For example,

(#$isa #$BarackObama #$UnitedStatesPresident)

(#$genls #$Mammal #$Animal)

or

Page 33: Artificial  Intelligence and  Computing Science

The same language gives rules for deducingnew knowledge:

(#$implies (#$and

(#$isa ?OBJ ?SUBSET) (#$genls ?SUBSET ?SUPERSET))

(#$isa ?OBJ ?SUPERSET))

Page 34: Artificial  Intelligence and  Computing Science

Intangible Things are things that are not physical -- are not made of, or encoded in, matter. These include events, like going to work, eating dinner, or shopping online. They also include ideas, like those expressed in a book or on a website. Not the physical books themselves, but the ideas expressed in those books. It is useful for a software application to know that something is intangible, so that it can avoid commonsense errors; like, for example, asking a user the color of next Tuesday's meeting.

What CYCcorp says CYC knows about `intangible things’.

Page 35: Artificial  Intelligence and  Computing Science

Questions CYC couldn’t answer in 1994

What colour is the sky?

What shape is the Earth?

If it’s 20 km from Vancouver to Victoria, and 20 km from Victoria to Sydney, can Sydney be 400 km from Vancouver?

How old are you?

(Prof. Vaughan Pratt)

Page 36: Artificial  Intelligence and  Computing Science

Hofstadter: Fluid Analogies

Human beings can understand similes, such as

``Mr Pickwick is like Christmas’’

Page 37: Artificial  Intelligence and  Computing Science

Example:

Who is the Michelle Obama of Canada?

Page 38: Artificial  Intelligence and  Computing Science

Michaelle Jean, Governor-General

Page 39: Artificial  Intelligence and  Computing Science

Head of government

Spouse Spouse

Page 40: Artificial  Intelligence and  Computing Science

Head of State

Spouse Spouse

Page 41: Artificial  Intelligence and  Computing Science

One of Hofstadter’s approaches to solving theseproblems is `Copycat’, a collection of independentcompeting agents.

If efg becomes efw, what does ghi become?

If aabc becomes aabd, what does ijkk become?

Page 42: Artificial  Intelligence and  Computing Science

Inside Copycat:

ijkk

ij(kk)

ij(ll)

(ijk)k

(ijk)l

aabc:ijkk

aabd:hjkk

aabd:jjkk

Page 43: Artificial  Intelligence and  Computing Science

If efg becomes efw, what does ghi become?

If aabc becomes aabd, what does ijkk become?

COPYCAT suggests whi and ghw

COPYCAT suggests ijll and ijkl and jjkk and hjkk

Page 44: Artificial  Intelligence and  Computing Science

Hofstadter:

``What happens in the first 500 milliseconds?”

Page 45: Artificial  Intelligence and  Computing Science

Find the O

Page 46: Artificial  Intelligence and  Computing Science

X X X X X X X X X X X

X X X X X X X X X X X

X X X X X X X X X X X

X X X X X X X X O X X

X X X X X X X X X X X

Page 47: Artificial  Intelligence and  Computing Science

Find the X

Page 48: Artificial  Intelligence and  Computing Science

X X X X X X X X X X X

X X X X X X X X X X X

X X X X X X X X X X X

X X X X X X X X X X X

X X X X X X X X X X X

Page 49: Artificial  Intelligence and  Computing Science

Find the O

Page 50: Artificial  Intelligence and  Computing Science

X X O X X X O X X O X

X X X X X O X X O X X

X X X O X X O X O X X

X X O X X O X X O X X

O X X X X X O X X O X

Page 51: Artificial  Intelligence and  Computing Science

What eye sees What I see

Page 52: Artificial  Intelligence and  Computing Science

The Cutaneous Rabbit

Page 53: Artificial  Intelligence and  Computing Science

Naive model of perception:

World Vision Awareness

Page 54: Artificial  Intelligence and  Computing Science

World Vision and Knowledge Awareness

Better model of perception:

Page 55: Artificial  Intelligence and  Computing Science

We recognise allthese as instances of the letter `A’.

No computer can do this.

Page 56: Artificial  Intelligence and  Computing Science
Page 57: Artificial  Intelligence and  Computing Science

Hofstadter’s program `Letter Spirit’ attemptsto design a font.

Page 58: Artificial  Intelligence and  Computing Science

Naïve Physics

Page 59: Artificial  Intelligence and  Computing Science

Hayes, ‘Naïve Physics Manifesto’, 1978

“About an order of magnitude more work than anyprevious AI project …’’

Hayes, `Second Naïve Physics Manifesto’, 1985

“About two or three orders of magnitude morework than any previous AI project…”

Page 60: Artificial  Intelligence and  Computing Science

One sub-project of naïve physics:

Write down what an intelligent 10-year-oldknows about fluids

Page 61: Artificial  Intelligence and  Computing Science

Part of this is knowing how we talk about fluids:

For example:

Suppose Lake Chad dries up in the dryseason and comesback in the wetseason.

Is it the same lake

when it comes back?

Page 62: Artificial  Intelligence and  Computing Science

Suppose I buy a cup ofcoffee, drink it, thenget a free refill.

Is it the same cup ofcoffee after the refill?

Page 63: Artificial  Intelligence and  Computing Science

2011: IBM’s Watson Wins Jeopardy

Page 64: Artificial  Intelligence and  Computing Science

Inside Watson: 4 Terabytes disk storage: 200 million pages(including all of Wikipedia)

16 Terabytes of RAM

90 3.5-GHz eight-core processors

Page 65: Artificial  Intelligence and  Computing Science

One of the components of Watson is a Google-like searchalgorithm.

For example, a typical Jeopardy question in the category`American Presidents’ might be

``The father of his country, he didn’t really chop down acherry tree’’

Try typing `father country cherry tree’ into Google

The first hit is `George Washington – Wikipedia’

But Watson also needs to know how confident it shouldbe in its answers

Page 66: Artificial  Intelligence and  Computing Science

Conspicuous Failures, Invisible Successes

In 2012, we have nothing remotely comparable to 2001’s HAL.

On the other hand, some complex tasks, such as attaching a printer to a computer, have become trivially easy

Page 67: Artificial  Intelligence and  Computing Science

A different approach: robot intelligence

Grey Walter’s machina speculatrix, 1948

Page 68: Artificial  Intelligence and  Computing Science

BEAM robotics,

Queen Ant, a light-seeking hexapod, 2009

Page 69: Artificial  Intelligence and  Computing Science

AI Now: Robot Intelligence

Rodney Brooks, `Cambrian Intelligence’

-Complex behaviour can arise when a simple system interacts with a complex world.

-Intelligent behaviour does not require a symbolic representation of the world.

Page 70: Artificial  Intelligence and  Computing Science

SPIRIT: Two years onMars and still going.

Page 71: Artificial  Intelligence and  Computing Science

Brook’s approach invites us to reconsiderour definition of intelligence:

…is it the quality that distinguishesAlbert from Homer?

Page 72: Artificial  Intelligence and  Computing Science

…or the qualitythat distinguishesAlbert and Homerfrom a rock?

Page 73: Artificial  Intelligence and  Computing Science

`Chess is the touchstone of intellect’

-- Goethe

…but perhaps we are most impressed by just those of our mental processes that move slowly enough for us to notice them…

Page 74: Artificial  Intelligence and  Computing Science

Strong AI:

``We can build a machine that will have a mind.’’

Weak AI:

``We can build a machine that acts like ithas a mind.’’

Page 75: Artificial  Intelligence and  Computing Science

Strong AI (restatement):

``We can build a machine that, solely byvirtue of its manipulation of formal symbols, will have a mind.’’

Page 76: Artificial  Intelligence and  Computing Science

Hans Moravec:

``We will have humanlike competence in a $1,000 machine in about forty years.’’

---- 1998

Page 77: Artificial  Intelligence and  Computing Science

Hubert Dreyfus:

``No computer can ever pass the TuringTest, or do any of the following things[long list, including `play master-levelChess’].’’

1965; MacHack beat Dreyfus in 1967

Page 78: Artificial  Intelligence and  Computing Science

…and if a program did pass the Turing test, what then?

Page 79: Artificial  Intelligence and  Computing Science

John Searle:

``Even if a computer did pass the Turingtest, it would not be intelligent, as wecan see from the Chinese Room argument’’

Page 80: Artificial  Intelligence and  Computing Science
Page 81: Artificial  Intelligence and  Computing Science

John Bird:

``A computer is a deterministic system,and hence can have neither free-will,responsibility or intelligence -- whether it passes the Turing test or not.’’

Page 82: Artificial  Intelligence and  Computing Science

``This is an AND gate:

A

B

C

A B C

0011

0101

0001

Given A and B, does the computer have any choiceabout the value of C?

Page 83: Artificial  Intelligence and  Computing Science

… but a computer is just a collection of AND gates andsimilar components. If none of these components canmake a free choice, the computer cannot make a free choice.’’

Page 84: Artificial  Intelligence and  Computing Science

The Brazen Head