Turing Test & Intelligence Turing Test & Intelligence

Turing’s GoalTuring’s Goal

Alan Turing, Computing Machinery and Alan Turing, Computing Machinery and Intelligence, 1950:Intelligence, 1950: Can machines think?Can machines think? How could we tell?How could we tell?

Turing’s AgendaTuring’s Agenda““I propose to consider the question, ‘Can machines think?’ I propose to consider the question, ‘Can machines think?’ This should begin with definitions of the meaning of the terms This should begin with definitions of the meaning of the terms ‘machine’ and ‘think’. The definitions might be framed so as to ‘machine’ and ‘think’. The definitions might be framed so as to reflect so far as possible the normal use of the words, but this reflect so far as possible the normal use of the words, but this attitude is dangerous. If the meaning of the words ‘machine’ attitude is dangerous. If the meaning of the words ‘machine’ and ‘think’ are to be found by examining how they are and ‘think’ are to be found by examining how they are commonly used it is difficult to escape the conclusion that the commonly used it is difficult to escape the conclusion that the meaning and the answer to the question, ‘Can machines meaning and the answer to the question, ‘Can machines think?’ is to be sought in a statistical survey such as a Gallup think?’ is to be sought in a statistical survey such as a Gallup poll. But this is absurd. Instead of attempting such a definition I poll. But this is absurd. Instead of attempting such a definition I shall replace the question by another, which is closely related shall replace the question by another, which is closely related to it and is expressed in relatively unambiguous words.”to it and is expressed in relatively unambiguous words.”

— — Alan Turing, Computing machinery and intelligence, 1950Alan Turing, Computing machinery and intelligence, 1950

Turing’s “Imitation Game”Turing’s “Imitation Game”

Interrogator B (a person) A (a machine)

Necessary versus Necessary versus Sufficient ConditionsSufficient Conditions

Is ability to pass a Turing Test a necessary condition of Is ability to pass a Turing Test a necessary condition of intelligence?intelligence? ““May not machines carry out something which ought May not machines carry out something which ought

to be described as thinking but which is very different to be described as thinking but which is very different from what a man does? This objection is a very strong from what a man does? This objection is a very strong one, but at least we can say that if, nevertheless, a one, but at least we can say that if, nevertheless, a machine can be constructed to play the imitation machine can be constructed to play the imitation game satisfactorily, we need not be troubled by this game satisfactorily, we need not be troubled by this objection.” — Turing, 1950objection.” — Turing, 1950

Is ability to pass a Turing Test a sufficient condition of Is ability to pass a Turing Test a sufficient condition of intelligence?intelligence?

Turing: Replace the QuestionTuring: Replace the Question““Instead of attempting such a definition I shall replace the question by Instead of attempting such a definition I shall replace the question by another. another. ... The original question, ‘Can machines think?’ I believe to be ... The original question, ‘Can machines think?’ I believe to be too meaningless to deserve discussion.too meaningless to deserve discussion.””

— — Turing, 1950Turing, 1950

““If Turing intends that the question of the success of the machine at the If Turing intends that the question of the success of the machine at the imitation game replace the question about machines thinking, thenimitation game replace the question about machines thinking, then it is it is difficult to understand how we are to judge the propriety and adequacy difficult to understand how we are to judge the propriety and adequacy of the replacement if the question being replaced is too meaningless to of the replacement if the question being replaced is too meaningless to deserve discussiondeserve discussion. Our potential interest in the imitation game is . Our potential interest in the imitation game is aroused not by the fact that a computer might learn to play yet another aroused not by the fact that a computer might learn to play yet another game, but that in some way this test reveals a connection between game, but that in some way this test reveals a connection between possible computer activities and our ordinary concept of human possible computer activities and our ordinary concept of human thinking.”thinking.”

— — Moor, 1976Moor, 1976

Current ViewsCurrent Views

The Turing Test is not a sufficient condition The Turing Test is not a sufficient condition for intelligence.for intelligence. Davidson Davidson semantics semantics Searle Searle intentionality intentionality Block Block richness of information processing richness of information processing

The Turing Test is a sufficient condition for The Turing Test is a sufficient condition for intelligence.intelligence. DennettDennett

Dennett: The Turing Test is SufficientDennett: The Turing Test is Sufficient

““The Turing test in unadulterated, unrestricted form, as Turing The Turing test in unadulterated, unrestricted form, as Turing

presented it, is plenty strong if well used. I am confident that no presented it, is plenty strong if well used. I am confident that no

computer in the next twenty years is going to pass the unrestricted computer in the next twenty years is going to pass the unrestricted

Turing test. They may well win the World Chess Championship or Turing test. They may well win the World Chess Championship or

even a Nobel Prize in physics, but they won't pass the unrestricted even a Nobel Prize in physics, but they won't pass the unrestricted

Turing test. Nevertheless, it is not, I think, impossible in principle Turing test. Nevertheless, it is not, I think, impossible in principle

for a computer to pass the test, fair and square. I'm not running for a computer to pass the test, fair and square. I'm not running

one of those a priori ‘computers can't think’ arguments. one of those a priori ‘computers can't think’ arguments. I stand I stand

unabashedly ready, moreover, to declare that any computer that unabashedly ready, moreover, to declare that any computer that

actually passes the unrestricted Turing test will be, in every actually passes the unrestricted Turing test will be, in every

theoretically interesting sense, a thinking thingtheoretically interesting sense, a thinking thing..””

— — Dennett, 1985Dennett, 1985

I conclude that the capacity to emit sensible responses I conclude that the capacity to emit sensible responses is not sufficient for intelligence, and so the is not sufficient for intelligence, and so the neo-Turing neo-Turing Test conception of intelligence is refutedTest conception of intelligence is refuted (along with (along with the older and cruder Turing Test conceptions). I also the older and cruder Turing Test conceptions). I also conclude that whether behavior is intelligent behavior conclude that whether behavior is intelligent behavior is in part a matter of how it is produced. Even if a is in part a matter of how it is produced. Even if a system has the actual and potential behavior system has the actual and potential behavior characteristic of an intelligent being, if its internal characteristic of an intelligent being, if its internal processes are like those of the machine described, it is processes are like those of the machine described, it is not intelligent.not intelligent.

— — Block, 1981Block, 1981

Block: The Turing Test is InsufficientBlock: The Turing Test is Insufficient

The Turing SyllogismThe Turing Syllogism

If an agent passes a Turing Test, then it If an agent passes a Turing Test, then it produces a sensible sequence of verbal produces a sensible sequence of verbal responses to a sequence of verbal stimuli.responses to a sequence of verbal stimuli.

If an agent produces a sensible sequence If an agent produces a sensible sequence of verbal responses to a sequence of of verbal responses to a sequence of verbal stimuli, then it is intelligent.verbal stimuli, then it is intelligent.

Therefore, if an agent passes a Turing Therefore, if an agent passes a Turing Test, then it is intelligent.Test, then it is intelligent.

The Occasional ConceptionThe Occasional Conception

If an agent produces a If an agent produces a sensible sequence of sensible sequence of verbal responses to a verbal responses to a sequence of verbal sequence of verbal stimuli, then it is stimuli, then it is intelligent. intelligent.

If an agent hasIf an agent has the capacitythe capacity to produce a to produce a sensible sequence of verbal responses to a sensible sequence of verbal responses to a sequence of verbal stimuli,sequence of verbal stimuli, whatever they whatever they may bemay be, , then it is intelligent.then it is intelligent.

The Capacity ConceptionThe Capacity Conception

Memorizing all possible answers?Memorizing all possible answers?(Bertha’s Machine)(Bertha’s Machine)

Exponential GrowthExponential Growth

Assume each time the judge asks a question, Assume each time the judge asks a question, she picks between two questions based on what she picks between two questions based on what has happened so farhas happened so far

Questions Asked Possible responsesQuestions Asked Possible responses

11 2222 4433 8844 161655 323266 6464n n 22nn

Storage versus LengthStorage versus Length

exponentialexponentialexponentialexponential

n=10n=10 n=20n=20 n=30n=30 n=40n=40 n=50n=50 n=60n=60

nn .00001.00001secondsecond

.00002.00002secondsecond

.00003.00003secondsecond

.00004.00004secondsecond

.00005.00005secondsecond

.00006.00006secondsecond

22nn .001.001secondsecond

1.01.0secondsecond

17.917.9minutesminutes

12.712.7daysdays

35.735.7yearsyears

336336centuriescenturies

33nn .059.059secondsecond

5858minutesminutes

6.56.5yearsyears

38553855centuriescenturies

2x102x1088

centuriescenturies1.3x101.3x101313

centuriescenturies

(one algorithm step = 1 microsecond)

(Figure 1.2 from Garvey & Johnson 1979)

Polynomial vs. exponential time complexity

The Compact ConceptionThe Compact Conception

If an agent has the If an agent has the capacitycapacity to produce a to produce a sensible sequence of verbal responses to sensible sequence of verbal responses to an an arbitrary sequencearbitrary sequence of verbal stimuli of verbal stimuli without requiring exponential storagewithout requiring exponential storage, then , then it is intelligent.it is intelligent.

Size of the UniverseSize of the UniverseHere, now

Big bang

15*109 light-yearsT

ime

Storage Capacity of the UniverseStorage Capacity of the Universe

Volume:Volume: (15*10 (15*1099 light-years) light-years)33 = =

(15*10(15*1099*10*101616 meters) meters)33

Density:Density: 1 bit per (10 1 bit per (10-35-35 meters) meters)33

Total storage capacity:Total storage capacity: 10 10184184 bits bits < < 1010200200 bits bits < < 22670670 bits bits

Critical Turing Test length:Critical Turing Test length: 670 bits 670 bits << 670 characters 670 characters < < 140 words < 140 words < 1 minute1 minute

The universe is not big enough to hold a bertha machine

The universe is not big enough to hold a bertha machine