There are three sections to this
book. The first, ‘Precursors’, includes antecedents to the Turing Test. As is
well known, Descartes held the view that ultimately what distinguished humans
from animals to an external observer is language use. By providing a selection
from Discourse on Method and ‘Letter
to the Marquess of Newcastle’, Shieber provides evidence that Descartes and
Turing might have agreed on the sufficiency of passing the Turing test for
establishing intelligence, or the presence of a soul. By having these readings
next to one another, one notices very clearly where they might have differed.
Descartes rationalism and Turing’s empiricism provide the basis for their
differing positions over whether any machine might pass the Turing Test. For
example, Turing muses over the tabula
rasa nature of newborn children, and suggest that this might guide
engineering decisions for making learning machines. "Presumably the child-brain
is something like a note-book as one buys it from the stationers. Rather little
mechanism, and lots of blank sheets" says Turing (Turing, Shieber 90-91).
Of course, Descartes holds that the engine of language use is the rational
soul, not merely "€¦lodged in the human body like a helmsman in his ship,
except perhaps to move its limbs, but that it must be more closely joined and
united with the body€¦ when we know how much the beasts differ from us, we
understand much better the arguments which prove that our soul is of a nature
entirely independent of the body€¦" (Descartes, Shieber 29). Curious in the
Precursors section is the inclusion of a selection from de La Mettrie’s Machine Man. While of interest to those
concerned with animal cognition, say, it is only of relevance to the Turing
Test construed as a necessary condition on intelligence. Shieber shows us he is
aware that Turing only thought of his test as a sufficient condition. However,
he, as do many authors in this text, seems to lapse into discussions of its
necessity, which often distracts from more fruitful endeavors.

Shieber
himself introduces many of the articles, and each of the three sections with
commentary. The portion of his commentary I found most interesting and helpful
is part of an introduction to a subsection, ‘Ephemera’, to the second main
section of the volume, ‘Turing’s Test’. In a portion with the heading ‘Sex and
the Turing Test’, Shieber reviews some of the published positions according to
which the traditional notion of the Turing Test misunderstands Turing’s
intentions in his original Mind
article. Famously, considers the gendered ‘Imitation Game’, and then
ambiguously asks: ‘what happens when a machine takes the place of the man’ (to
rephrase Turing, [Shieber 68])?  The imitation
game involves an attempt by a judge to determine on the basis of written
responses to queries which of two subjects is a man, and which is a woman.
Shieber cites interpreters according to whom the ‘machine imitation game’
includes a judge who thinks that he or she must detect which responder is a
woman and which is a man, but is in fact receiving responses from a woman and a
machine. The orthodox view, which Shieber supports, is that Turing intended
wider scope to the phrase ‘takes the place of’. The resulting Turing Test is an
attempt by a judge to decide which of two verbal respondents is a human and
which is a machine, with the knowledge that there is exactly one of each.
Shieber’s commentary deftly applies Turing’s included ‘Ephemera’, hitherto unpublished
in book form, to convincingly argue that the gendered interpretation is a
tortured misreading of an unfortunate ambiguity. He appeals to quotes of
Turing’s to make his point, such as this one.

I think it is probable for instance that at the end of
the century it will be possible to programme a machine to answer questions in
such a way that it will be extremely difficult to guess whether the answers are
being given by a man or by the machine. I am imagining something like a
viva-voce examination, but with the questions and answers all typewritten in
order that we not consider such irrelevant matters as the faithfulness with
which the human voice can be imitated. (Turing, Shieber 114)

This quote, taken from a BBC lecture the year following the
publication of Turing’s Mind article
is powerful evidence against the gendered variant. There are alternatives one
can pursue, however, other than arguing over precisely what Turing originally
meant. For example, Shieber doesn’t mention Sterrett’s argument (2000) that
even if Turing didn’t mean for us to pursue the gendered variant of the
imitation game, there is important value in considering it. Nevertheless, the
editor does a service in providing a textually supported, concise argument for
the view that Turing meant what we usually think he did in introducing what is
now called the Turing Test.

The third section contains a wide
selection of some of the most important reponses and counter-responses to
Turing’s Mind article. This volume
contains a hitherto unpublished, four page article by Noam Chomsky on the
significance and interest of the Turing Test. It is unclear whether this
article was written well in advance of the edited volume, solicited by Shieber,
or was written just for The Turing Test.
In any case, the copyright for the article is 2002 while that of the book is
2004. On the one hand, Chomsky notes that in formulating what is now called the
Turing Test, Turing had two goals: 
"constructing better machines [and] gaining insight into human
intelligence" (Chomsky, Shieber 317). He is quite critical of the Turing
Test and whether trying to build computers that would pass it would help us
pursue these two goals. Chomsky writes:

Questions about
computational-representational properties of the brain are interesting and it
seems important, and simulation might advance theoretical understanding. But
success in the imitation game in itself tells us nothing about these matters.
Perhaps, as Turing believed, the imitation game would provide a stimulus for
pursuit of the two "useful lines of research" he advocated; he said
little about why this research strategy is preferable to other ways to improve
machine capacity and study human intelligence, and it does not seem obvious,
apart from some cultural peculiarities that an outside observer might assess
with a critical eye. (319)

Like the rest of Chomsky’s article,
this selection seems hastily written and is unclear. What exactly ‘seems
important’ in the first sentence above? I assume Chomsky means something like
‘It is important that we try to answer questions about
computational-representational properties of the brain’. What ‘does not seem
obvious, apart from some cultural peculiarities€¦’ mean in the last sentence? I
can guess one of the following two things: the Turing Test might allow cultural
peculiarities of instances of artificial intelligence to be critically assessed
by observers who did not share those peculiarities; or, the idea that the
Turing test might contribute to the second research goal is itself a cultural
peculiarity that other scientists might not share. In my opinion, the remainder
of Chomsky’s article does not fare much better. However, future discussions of
the Turing Test will no doubt enlist the conclusions reached therein.

It is easy to criticize an edited
collection on any problem or issue in philosophy by pointing out selections
which ought to be have included but weren’t. To his credit, Shieber tells the
reader that he is omitting important lines of discussion and offers directions
for supplementing the book. However, even though he mentions the lengthy
discussion of the ‘Total Turing Test’, he only cites one of Harnad’s more
recent articles on the matter. To get a broader picture of the lively
contemporary debates about the Turing Test and important articles therein, I
refer the reader to Saygin et al.’s recent article ‘Turing Test: 50 Years
Later’ (2000) and the special issues of Minds
and Machines ‘The Turing Test: Past, Present, and Future’ in which it
appears. Shieber has selected what are some of the most influential and
interesting articles, however. A strength of the selections is the degree to
which subsequent authors, without doing so by name, take up lines of argument
presented by the others. One thread that goes through many articles is the
issue of algorithmic intelligence. Turing’s radio broadcast with a few
contemporary scientists (Chapter 7) and articles by Sampson, Purtill, Searle,
Block and Dennett return again and again to the issue of whether implementing a
mindless algorithm could possibly count as true intelligence, thus questioning
the sufficiency of the Turing Test.

As for articles that were included
but might have been better left out, there are few. Shieber introduces the
early article by Leonard Pinsky ‘Do Machines Think about Machines Thinking’ by
saying "It is hard to know what to make of this fairy tale, and probably
not sensible to make too much of it. But as the first in the long history of
Turing responses, it may serve to prevent our taking these issues too seriously"
(Shieber, 141). In a nutshell, Pinsky’s short article uses the Turing Test as a
quick route to a reductio of the late Wittgensteinean position of therapeutic
positivism. It is clear by reading the article that Pinsky has little interest
in the Test itself, merely employing it to his own ends. Priority in the
literature for mentioning the Turing Test doesn’t seem to warrant inclusion in
an otherwise well selected set of articles. On the other hand, Robert French’s
famous article on subcognitive performance in the Turing Test stands as
somewhat of an island in the book, in the absence of other articles responding
to its arguments and claims. Overall, the editing and presentation of the
articles is superb, although the reader may be a little surprised to see that Philosophy and the Mirror of Nature is
listed both in the bibliography of the book as a whole, and one following an
article, as having been written by Amélie Rorty, rather than Richard.

References

Saygin, P.A., Cicekli, I, and Akman, V., 2000. Turing Test:
50 Years Later. Minds and Machines
Vol. 10 No. 4 pp. 463-518.

Sterret, Susan G., 2000. Turing’s Two Tests for
Intelligence. Minds and Machines Vol.
10 No. 4 pp 541-559.

© 2004 Darren Abramson

Darren Abramson is a Ph.D. candidate in philosophy and
cognitive science at Indiana University. He holds a MSc in computer science
from Indiana, as well as a BA in philosophy from University of Toronto.

Categories: Philosophical, Psychology