Page 345 - Data Science Algorithms in a Week

326 Bert Olivier

simultaneously taking affective states into account, lest a caricature of the mind emerge,
which appears to be what mainstream AI-research has allowed to happen.
Such circumspect perspicacity does not sit well with the majority of other researchers in
the field, who generally do not merely set the question of the body aside, like Turing did
(because he realised its intractability), but simply ignore it, in the naïve belief that one can
legitimately equate the mind with software and the brain with hardware. This seems to imply,
for unreflective AI-developers, that, like software, human minds will, in future, be
“downloadable” to computers, and moreover, that human brains will – like computer
hardware – become “almost infinitely upgradable”. Anyone familiar with the phenomenology
of human beings, specifically of the human body, will know that this is a hopelessly naïve,
uninformed view. Take this passage from Merleau-Ponty, for instance, which emphasises the
embodied character of subjectivity (the “I”) as well as the reciprocity between human subject
and world (1962: 408):

I understand the world because there are for me things near and far, foregrounds and
horizons, and because in this way it forms a picture and acquires significance before me,
and this finally is because I am situated in it and it understands me.…If the subject is in a
situation, even if he is no more than a possibility of situations, this is because he forces
his ipseity into reality only by actually being a body, and entering the world through that
body…the subject that I am, when taken concretely, is inseparable from this body and
this world.

Mainstream AI-research’s reduction of the embodied human subject to
‘hardware/brain with software/mind’ rules out, from the start, grasping what is distinctive
about human beings – under the sway of the mesmerizing image of the computer, it
follows the heuristic path of reduction of what is complex to what is merely complicated,
and deliberately erases all indications that human or mental complexity has been elided.
It is clear that, unlike most of his mainstream colleagues, however, Gelernter is not in
thrall to the power of computers; from the above it is apparent that he is far more – and
appropriately so – under the impression of the complexity and the multi-faceted nature of
the human mind. His work raises the question (and the challenge to mainstream
‘computationalism’), whether AI-research can evolve to the point where it can produce a
truly human simulation of mind across the full spectrum of its functions (Olivier 2008),
instead of the reductive version currently in vogue.

SHERRY TURKLE ON THE ROBOTIC TURN

Sherry Turkle takes Gelernter’s assessment, that mainstream AI-research is
misguided because of its partial, ultimately reductive, ‘computationalist’ conception of
the human mind, to a different level in her book, Alone Together (2010). As I shall argue
below, it is not so much a matter of Turkle contradicting Gelernter when she elaborates