Tuesday, December 20, 2011


When you play Need For Speed on Playstation, there exists a mode of play in the game in which you, the player, chase a 'ghost car' around the track.  The ghost car is driving an optimal path around the circuit.  Your objective is to mirror the ghost car and learn the best 'line' through that level.  Let me draw a comparison to a momma bear and a cub and tie it all into Mechatron AI.

'Teaching' a mechanical being is the stuff of science fiction, right?  What is teaching, in the context of a mess of sensors, wires, microcontrollers running programs?  In the same vein, what is learning?  I have observed in the world of Artificial Intelligence a trend.  The trend being numerous attempts to 'create' an AI personality.  If we've learned our lessons from Wargames, then we know this being must be taught, it is only a child to the world.

A momma bear must teach her cub how to hunt, defend herself, how to avoid dangers of life in a forest, etc.  Building the bear cub's mind within a digital world is our first step into Mechatron AI.  What we need to develop is a blank slate that is capable of learning, but the lessons need to be taught or programmed.  I hate to use the word programmed though.  I feel it takes the life out of the experiment.  Let's coin a phrase.  What would be the best description to describe an MT receiving a set of instructions, in which it gets to decide the 'best' course of action, as IT sees fit?  Free Will Code....  Will Free Code....  Will Code For Food...wait....


YES.  Electronic DNA.

So what is eNa?  You tell me.  Ok, eNa is the 'helper' code that is 'soul transferred' to the MT.  No more Uploads or Downloads, those are for inanimate objects!  More soon.

No comments:

Post a Comment