Deliberative Algorithms as Scaffolding

In this sequence, we will investigate the framing of deliberative cognitive algorithms as a form of scaffolding for black box (particularly generative) models.

In particular, we will consider rationality techniques implemented by the conscious mind as scaffolding for the unconscious mind. Artificial intelligence researchers seek to build intelligent machines, cognitive scientists study intelligent behavior in humans, and decision theorists study optimal decision making. But how can we approach the principled construction of rationality enhancing cognitive techniques for humans, without read or write access to much of the psyche? By making optimal use of black box access. 

(Banner image from AlphaGo paper).