Wednesday, November 4, 2015

Self knowledge in Asa H

A robot embodied Asa H has inputs from its physical senses.  On the lowest level in the concept/memory hierarchy Asa feels things like its level of battery charge, temperature, pain/damage, sight, sound, touch, acceleration, etc.
Asa can also have access to internal/software features. On the various levels in the concept hierarchy it can accept as input things like the size of the current casebase, the current learning rate ("L"), how often it is attempting case extrapolation ("skip"), etc.  (see my blog of  10 Feb. 2011 for an example of "L" and "skip" )  We can also measure,  record, and input the time spent in any of Asa's algorithmic processes. Asa can then learn to adjust/optimize any of these quantities. (See my book Twelve Papers, chapter 1, page 15, self monitoring) In this way Asa can sense its own thought processes. Is this the nature of qualia?

No comments:

Post a Comment