Create a simulation with artificial creatures evolving in a virtual environment that rewards evolving more and higher intelligence (though need to define "intelligence" which may be tricky). Can we create a general AI by simply running that evolution simulation at high speed?
What is the difference between non-intelligently adapting to an environment and intelligently understanding an environment and acting according to understanding? Both succeed in exploiting the environment to maximize benefit. Intelligence probably adapts quicker to change, not taking generations to discover the best adaptation.
Exactly how should the simulation work? How can it reward intelligence?
No comments :
Post a Comment