Cog is a robot with an upper-torso humanoid body that learns to interact with people through various “senses” (Figure 8.1). It is different in many respects from any other artifact that we have seen so far. In the vocabulary used by its designers, Cog is:

• embodied – it has a body (or some fraction thereof) similar to the human body;

• embedded – it is “socialized” (in a very minor way);

• developing – there is a “baby” version of it, Kismet, that is meant to go through development;

• integrated – it is equipped with, and is meant to integrate the data from, the equivalents of various sensory organs.

Furthermore, Cog is different from a typical robot in being “general-purpose” – it is meant to become the dream household robot that could act as a pet, a baby, a servant, and, ultimately, as a companion. “The distinction between us and robots is going to disappear” (Brooks 2002a: 236).

Even this briefest of introductions to Cog seemingly points to a radical departure from everything that has heretofore happened in AI. Classical AI, we should recall, views cognition as abstract (physical embodiment is irrelevant), individual (the solitary mind is the essential locus of intelligence), rational (reasoning is paradigmatic of intelligence), and detached (thinking is separated from perception and action) (Smith 1999).