6 Preliminaries
However, this evolutionary type of functionalist argument cuts both ways: first,
it exposes a fallacy in the claim made by AI critics that a disanalogy in structure,
that is in the way in which and the means by which some task is accomplished,
precludes analogy in function. This is basically the inverse of Turing Machine
functionalism, or indeed of any functionalist argument of this kind, and it appears
to be the bottom line of John Searle’s AI critique in (1980) and many other critiques that follow his route (which include philosophers as eminent as Dretske
1985 and Fodor 1981). If similarity in structure were thus required, and unless
that similarity were superficial or coincidental, the requirement would be one of
homology. This is a condition that, by definition, any machine would invariably
fail at. Moreover, it is a condition that does not respond to the claim of functional
analogy, as homology and analogy are independent affairs. Second, and for the
same reason, similarity in structure is uninformative as to an analogy of function.
Hence, human-likeness of some machine in appearance or behaviour does not
warrant any inference with respect to intellectual abilities of that machine.
Despite this implication of a functionalist argument, and in keeping with the
questionable interpretation of Turing’s imitation game as a test for, or definition
of, intelligence, AI was long committed to the criterion of human-likeness of its
systems, in appearance and behaviour. This commitment is reflected in the definition of the research programme that kept dominating the field at least until the
1990s: AI was understood as an inquiry into the nature of the human mind in
which theories about its structure, properties and functions were tested by means
of computer programmes, computer systems or robots as models. However, the
long-standing dominance of AI as the modelling and simulation of human thought
processes may have helped to obscure rather than illuminate other, possibly more
instructive, roles of computers and robots as models concerning (if not to say of )
human cognition and action.
Turing himself deliberately styled his inquiries into the possibility of thinking
machines as flights of fancy that he only could – and did – wish to be true, but
he also pointed towards two implications of the presence and use of computing
machinery that is much closer to home in many respects:
The original question, ‘Can machines think?’ I believe to be too meaningless
to deserve discussion. Nevertheless I believe that at the end of the century
the use of words and general educated opinion will have altered so much
that one will be able to speak of machines thinking without expecting to be
contradicted.
(Turing 1950, 442)
Irrespective of the question of whether they match human intellectual abilities,
the accomplishments of computing machinery would ultimately alter our understanding of what human thinking is – a point made so explicit by Turing that
I keep wondering why it has not raised more scholarly attention. Digital computers may be relevant to human cognition, and to how we conceive of it, in other
ways than simulating human thought processes.
However, this evolutionary type of functionalist argument cuts both ways: first,
it exposes a fallacy in the claim made by AI critics that a disanalogy in structure,
that is in the way in which and the means by which some task is accomplished,
precludes analogy in function. This is basically the inverse of Turing Machine
functionalism, or indeed of any functionalist argument of this kind, and it appears
to be the bottom line of John Searle’s AI critique in (1980) and many other critiques that follow his route (which include philosophers as eminent as Dretske
1985 and Fodor 1981). If similarity in structure were thus required, and unless
that similarity were superficial or coincidental, the requirement would be one of
homology. This is a condition that, by definition, any machine would invariably
fail at. Moreover, it is a condition that does not respond to the claim of functional
analogy, as homology and analogy are independent affairs. Second, and for the
same reason, similarity in structure is uninformative as to an analogy of function.
Hence, human-likeness of some machine in appearance or behaviour does not
warrant any inference with respect to intellectual abilities of that machine.
Despite this implication of a functionalist argument, and in keeping with the
questionable interpretation of Turing’s imitation game as a test for, or definition
of, intelligence, AI was long committed to the criterion of human-likeness of its
systems, in appearance and behaviour. This commitment is reflected in the definition of the research programme that kept dominating the field at least until the
1990s: AI was understood as an inquiry into the nature of the human mind in
which theories about its structure, properties and functions were tested by means
of computer programmes, computer systems or robots as models. However, the
long-standing dominance of AI as the modelling and simulation of human thought
processes may have helped to obscure rather than illuminate other, possibly more
instructive, roles of computers and robots as models concerning (if not to say of )
human cognition and action.
Turing himself deliberately styled his inquiries into the possibility of thinking
machines as flights of fancy that he only could – and did – wish to be true, but
he also pointed towards two implications of the presence and use of computing
machinery that is much closer to home in many respects:
The original question, ‘Can machines think?’ I believe to be too meaningless
to deserve discussion. Nevertheless I believe that at the end of the century
the use of words and general educated opinion will have altered so much
that one will be able to speak of machines thinking without expecting to be
contradicted.
(Turing 1950, 442)
Irrespective of the question of whether they match human intellectual abilities,
the accomplishments of computing machinery would ultimately alter our understanding of what human thinking is – a point made so explicit by Turing that
I keep wondering why it has not raised more scholarly attention. Digital computers may be relevant to human cognition, and to how we conceive of it, in other
ways than simulating human thought processes.