all systems learn in some fashion. They may not learn in the
   manner we think of as learning but it is still learning.

   Learning requires elasticity to form within a system and
   hysteresis. Hysteresis is when all of the past inputs bear upon
   the results provided by the current input via a feed-forward
   loop.

   In short, there are two states simultaneously:
   There is the "culmination of all past inputs"
   then there is
   "new input".

   But the new input does not work alone. All of the past bears
   upon the present, while simultaneously the present affects the
   culmination of all past inputs.

   This results in a lag because of processing time.

   This lag is working memory and consciousness.

   The "two states" are the elasticity; it "wants" for the lack of
   a better word, to return to the cultimation of past inputs.

   There is always a slight difference. Within that difference is a
   form of choice.

   A thermostat has a form of free will.
   But that free will is within constraints.

   Our free will is ALSO within constraints.

   An AIs free will will ALSO be within constraints.