AGI architecture measures human parameters and optimizes human performance

Abstract

AGI could manifest itself in human-computer interactions. However, the computer should know what is on the mind of the user, since reinforcement learning, the main building block of AGI, is severely spoiled for partially observed states. Technological advances offer tools to uncover some of these hidden components of the 'state'. Here, for the first time, we apply an AGI architecture for the optimization of human performance. In particular, we measure facial parameters and optimize users' writing speed working with head motion controlled writing tool. We elaborate on how to extend this optimization scheme to more complex scenarios. © 2011 Springer-Verlag Berlin Heidelberg.