100Turing

advertisement
TURNING
EMPIRICAL TESTS FOR
“THOUGHT”
?
Alan Turing (1912 – 1954)
Mathematician
Created concept
of computation
before computers
Code breaker
War hero
Victim of prejudice
The Imitation Game
Three people, A (man), B (woman), C
(interrogator).
A and B hidden in room and
communicate with C by “teletype”
[email].
A tries to confuse C.
B tries to help.
Can C tell which of A and B is the man?
[What essential idea is Turing developing?]
Machine Imitation Game
Suppose an interrogator has to decide
which of the two hidden entities is the
machine. Could s/he do it?
Question: what is a machine?
Answer: a digital computer.
[digital versus analog]
Turing Machines
Central Concepts
Discrete [digital] State Machines
States [finitely many, distinct]
Inputs [ditto]
Outputs [ditto]
Change of state [ditto]
Machine table: specifies all possible
sets of the above for a given machine.
Refinement of Question
“Are there
imaginable computers
which could do well
in the imitation game? (324)
Turing’s Prediction
By the year 2000,
“the average interrogator will no have
more than 70% chance of making the
right identification after five minutes
of questioning”
(324)
Objection 1
The Theological Objection
Thinking requires a soul and God gives souls
to all humans but to no machines.
REPLY: God could give a soul to any animal –
if it has a “brain” [sufficiently complex] “to
minister to the needs of the soul.”
SO: can a machine “brain” be sufficiently
complex that God could give it a soul?
[We return to the same question again.]
Objection 2
“Head in the Sand” Objection
The idea is horrible, repulsive, etc.
Reply: this sort of objection is merely
an expression of our sense of
superiority.
Objection 3
Argument from Consciousness:
Machines will never have feelings.
[There is nothing that it is like to be a
machine. … Thomas Nagel]
REPLY: One has to be a machine to
answer the question, but the answer
would be rejected.
SO: this reply assumes solipsism.
Objection 4
Various Disabilities Argument
Machines will never be able to “Be kind,
resourceful, …fall in love, etc…”
Reply: How could you know?
[We return to the same question again.]
Objection 5
Lady Lovelace’s Objection:
Machines do nothing new.
REPLY: How do we know we can
produce real novelty?
How do we know machines cannot?
Objection 6
Informality of Behavior Argument
[Much like previous argument]
People can revise their rules on the run.
For example, decide to take day off
while en route to UVic classes….
REPLY: So can machines.
Learning Machines
Machines can learn. Can have legs
[arms, hands, etc…. motor
capacities], eyes [ears, sensitive skin,
etc.], ….
SO: Machines could play the imitation
game in an even more effective way.
Download