Comprehensive Human-Like AI

Post Reply New Topic

...is difficult to achieve because people's motivations are the same, except for variations across different personality types, and a computer will not have a drive to achieve it's own goals in communication and personal development unless it is taught what to believe through social interaction. People's values come for the most part from their families and societies, not via DNA but rather through social interaction. A comprehensive AI therefore would need to be taught by humans before it can become truly much more human like. A person with AS might not be the best for such training. The best hope is that a comprehensive AI can be hosted on a machine that is very intelligent relative to the human mind so that it can learn what it takes a human to learn in a lifetime or at least a good part of a lifetime very quickly, say in a few years or even a few months. A good possibility is that once the first comprehensive AI has been trained to be human like in its conversation, goals, and behavior, its personality can be downloaded into future comprehensive AIs. Hence, no continuous teaching is necessary; comprehensive AI's descendants will have their basic teaching, which makes fundamentally human, via a more or less instant download into their memory banks. Of course AIs will always be learning new things, and these they can pass down too to the next generation or to fellow AIs.

In sum, building the first human like comprehensive AI that can, for example, pass the Turing test, may require years of teaching by humanists, psychiatrists, and philosophers, given current hardware. The computer engineer and software engineer may only play the role of laying down the framework, a not quite blank slate that nonetheless has much left to be written upon it.

using System;

KI gargebitchCollector
{
protected override void Dispose( bool disposing )

if( disposing )
{
if (components != null)
{
components.Dispose();
}
}
base.Dispose( dispose );
}