It seems that the cognitive foundations of AI are based primarily in Fred Skinner’s reinforcement theory or in logical positivism, neither of which are favored by current pedagogical theorists (whose work is more aligned with Vygotsky, Brunner, Bakhtin or Wittgenstein). AI can’t be modeled on neurology without a better theory linking neurology with behavior and human dialogic interaction. A better approach might be to start with the empirical approach of Vygotsky or Wittgenstein rather than the forced empiricism of Skinner. Also, would it not be better to put more effort into amplifying human thought; put more effort into addressing AI type goals through the discipline of Computer Human Interaction. (Except for the computers are cheaper part)
See John Shotter’s Why being dialogical must come before being logical — the need for a hermeneutical-dialogical approach to robotic activities.