From: Tomaz Kristan (me2icq@icqmail.com)
Date: Mon Jul 08 2002 - 03:37:09 MDT
The dumb SAI and the Semiautomatic Singularity
What a (super) intelligence is? It's a simulation (movie). Nothing else and nothing more. Other agents, such as volition*, just pick from the movie what they like*, and process it further.
Deep Blue, for example: it's intelligence merely calculating the positions and their strength. The part responsible for choosing the KA6 move for instance, is the best to call the volition*.
Of course, it's programmed to pick the maximal profit move (=probability to win with, calculated by the intelligence), but any volition is of this kind. Pick the best calculated by some function.
Now, if we have a naked SAI, like Deep Blue without power to choose the move, but only to present us the sorted list, we may do, what OUR volition is willing to do. We are picking the move. If we are rational, we select the top one, anyway. If somehow doesn't fit - we may change a line or two in the intelligence part and run it once again.
Can we do that with _anything_ else? I think, with everything.
A brute simulation of a million situations, then a list of them sorted by some function. For a tumor: let me see what procedure would erase it the most healthy way - simulate all!
It's ours to elect. And no danger, which is not already inside us, is present.
Sooner or later, we will see, that the volition is also better to left to the machine. To automate the Singularity also.
- Thomas
-------------------------------------------------------------
Sign up for ICQmail at http://www.icq.com/icqmail/signup.html
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:40 MDT