Ben said regarding an AI that could learn and play, without specific coding,
high-level chess, go, etc.

>Such a software system would be VASTLY superior to any existing AI software
>system.  But if it could do nothing else, it would still be terribly
>overspecialized and narrow compared to a human...

>I would see such a system as halfway between Deep Blue and humans, in terms
>of general intelligence.

Agreed.  But I would then confidently conclude that we were on the steep
part of the curve toward AGI, and that the challenge of creating
intelligence had been met.  This would be a difficult AGI test that is also
clear and simple to administer.  Much more telling than the Turing imitation
game test I think.

Hmmm, here is another test idea:  given a $10,000 budget, conceive and
execute a plan for a web-based software services business that will legally
return $30,000 profit within a year.  Closer to a true AGI test?

Later . . . Kevin C.

-------
To unsubscribe, change your address, or temporarily deactivate your subscription, 
please go to http://v2.listbox.com/member/?[EMAIL PROTECTED]

Reply via email to