Page 354 - Data Science Algorithms in a Week

Explorations of the ‘Transhuman’ Dimension of Artificial Intelligence 335

a being, constantly ‘replenished’ by accumulating, multiplicitous layers of information or
‘memory’, while every successive generation of organisms, according to Sheldrake,
inherits the collective memory from the generation before it, or contemporaneous to it, in
other parts of the world.
But is this ostensible resemblance between a certain kind of artificial intelligence (the
fictional, but informationally possible, idoru) and humans adequate to establish an
identity? Probably not – even if the analogy between the growing informational
‘foundation’ of which the idoru is the epiphenomenon, and generations of humans (as
beings that rely on ‘morphic resonance’ for ‘information’ regarding appropriate modes of
behaviour) is tenable, the difference would be precisely the uniqueness of every finite,
embodied human subject, compared to the transhuman, infinitely escalating aggregate of
information – vast as it already is – which might manifest itself in different forms, such
as the fictional, and transhuman, idoru.

CONCLUSIONS

What have the preceding reflections on manifestations of the transhuman in artificial
intelligence research brought to light? The brief examination of Jonze’s Her served the
important objective, to provide a kind of paradigmatic instance of what a transhuman AI
would be like, that is, what would make such a being recognisably ‘transhuman’ in its
virtually incomprehensible otherness. This fictional excursion prepared the way for a
brief consideration of David Gelernter’s contention, that when the human mind is
conceived of in terms of a ‘spectrum’ of mental functions covering rational thinking,
daydreaming, fantasy, free association as well as dreaming, the concentration of
mainstream AI-research on exclusively the first of these levels (as a model for AI),
appears to be seriously flawed. The image of AI that emerges from such
‘computationalist’ research would be truly ‘transhuman’. It was argued further that
Sherry Turkle’s work complements Gelernter’s through her foregrounding of the
irreducible differences between performatively impressive, intelligent and quasi-
affectionate androids (robots) and human beings: unlike humans, the former lack a
personal history. Christopher Johnson’s work, in turn, was shown as focusing on the
conditions of engineering AI in the guise of robots that would be convincing simulations
of human beings. Johnson finds in replication through ‘reverse-engineering’ the promise
of successfully constructing such robots. However, his reminder, that human beings are
distinguished by their uniqueness, implies that the difference between a transhuman,
‘neuromorphically’ engineered android and a human being would remain irreducible.
Returning to fiction, William Gibson’s perspicacious exploration of the potential for
artificial intelligence, harboured within the ever-expanding virtual realm of (digital)
information, was used to demonstrate its similarity with successive generations of human