steiny@scc.UUCP (Don Steiny) (06/04/84)
***
"Could a machine think?"
...
"Can a machine have a toothache?"
...
"You will certainly be inclined to say: 'A machine
can't have a toothache. A I will do now is to draw your
attention to to the use you have made of the word 'can'
and to ask you: 'Did you mean to say that all our
past experience has shown that a machine has never had
a toothache?' The impossibility of which you speak is a logical
one. The question is: What is the relation between thinking (or toothache)
and the subject which thinks, has toothache, etc.?
Wittgenstein, Ludwig
The Blue and the Brown Books.
"Preliminary Studies for the 'Philosophical
Investigations."
Copyright Basil Blackwell, 1958
Harper Colophon Books,
printed: 1960
p. 16
" ... 'Is it possible for a machine to think?'
(whether the action of this machine can be described and
predicted by the laws of physics or, possibly, only by laws of
a different kind applying to the behaviour of organisms).
And the trouble which is expressed in this question is not
really that we don't yet know a machine which could do the job.
The question is not analogous to to that which someone
might have had a hundred years ago: 'Can a machine liquify
a gas?' The trouble is rather that the sentence: 'A
mchines thinks (perceives, wishes)': seems somehow
ninsensical. It is as though we had asked 'Has the
number 3 a colour?' ('What colour could it be, as it
obviously has non of the colours known to us?') For in
one aspect of the matter, personal experience, far from
being the *product* of physical, chemical, physiological processes,
seems to be the very *basis* of all that we say with any nse
about such processes. Looking at in this way we are inclined
to use our idea of a building-material in yet another misleading
way, and to say that the whole world, mental and physical,
is made up of one material only.
ibid. pp. 47-48
Wittgenstein points out that because we use the
word 'thought' in the same way that we use the word
'table' as in:
1) She has a thought.
2) She has a table.
We are misled into believing that there is a 'thing'
called a 'thought' that we can somehow point to. The logical
extention of this is to imagine that the word 'thought'
refers to certain human behavior. Turing reframed
the question (approximately) as, can a machine exhibit
behavior that, were a human to exhibit that behavior,
we would explain the behavior as "thinking".
There is a view of language expoused by the school
of philosophy called the "logical positivists". This school
believes that words are signs that refer to things. Words
are labels for things.
Wittgenstein did not believe words were labels for
things. He believed that language was *activity* we engage in.
He said that language was a singaling function. This can be
seen in the iterpretation of 1 & 2 above.
1) She has a thought.
positivist view:
she - refers to some specific person we both know.
has - this person owns, posseses, is attributed with ...
a thought - some thing called a thought.
Wittgenstein's view:
Wittgenstein admonished to look for how a word was used.
A way #1 might be used is if there were several people. It might
be a signal that "she" wishes to take a turn speaking.
2)
positivist view:
same as #1
Wittgenstein's view:
I can generate several instances where #2 might be a signal
to affect future behavior of another. For instance, we might be
giving a party and trying to decide where to get an extra table.
If she does have a table, we do not have to call Bill, if she
does not have a table, we have to call Bill. #2 would be a
signal not to call Bill.
------------------
If we "look for how a word is used" for words like
"think", and "talk", we find that machines already do "think"
and "talk". Computers "talk" to busses and each other. If a
program is doing a massive computation, it is natural to say: "it's
thinking." Perhaps machines will ultimately "think" because
we will expand our defination of the word, instead of the
efforts of the AI community.
Don Steiny
Personetics
109 Torrey Pine Terr.
Santa Cruz, Calif. 95060
(408) 425-0382
ucbvax!hplabs!pesnta!scc!steiny
harpo!fortune!idsvax!scc!steiny