From: Justin Corwin (thesweetestdream@hotmail.com)
Date: Wed Mar 21 2001 - 18:38:11 MST
I personally would not want to be around if SIs start struggling for control
of something (like an antique Extro-5 shirt...;-)) and it goes violent.
Napoleon plus Genghis Kahn, plus all the technology you can imagine wouldn't
compare with the utter chaos that would result. and likely, people wouldn't
just sit by and let them duke it out, they'd take sides, adding to the chaos
and confusion.
i would hope most people would try to help the SysOp in whatever way they
could, adding their feathers of resources and smarts to the battle. But it
would be hard for them even to fathom the thrusts and parries of
superintelligent warfare.
>From: "Eliezer S. Yudkowsky" <sentience@pobox.com>
> >Justin Corwin wrote:
> >
> > 1, seed reactive nano-bots which will kill humans if tampered with, or
>if
> > some external action is taken by the Friendly SI, limiting the FSI's
> > actions.
> >
> > 2, sieze control of a non-essential bit of computronium, which
>unfortunately
> > contains sims, and hold *hostage* against FSI action.
> >
> > 3, blast any populated star system with hard gamma radiation, forcing
>FSI to
> > protect any sims there and waste resources.
The problem with my suggestions is that they're simple, and obviously not
superintelligent, i was just suggesting thrusts and strategies. threaten
citizens, sieze control of citizens, and force SysOp to counter anti-citizen
efforts that cost Adversary next to nothing. eh, i dislike thinking about
destructive things, it just seems thing lead inevitably to me doing so.
>
>If things get that far... but never mind. I agree with 1 and 3. I'm not
>quite sure about 2, since the action of the UFSI is predicated on being
>able to predict the reaction of the FSI, and the FSI might use the "I do
>not negotiate with terrorists logic" or a variation thereof... but never
>mind. The main point is that Gordon Worley seemed to think a group of
>cooperating Terran UFSIs would be a better protection for the Solar System
>at this point - would minimize total suffering - due to an UFSI's superior
>grasp of battle tactics, or an FSI's unwillingness to tolerate any
>casualties whatsoever, or something like that.
hm. i remember somebody proposing we control SI AIs by playing them off each
other, some application of game theory. but i don't think that would work to
well. and besides, ten big folk who's interest's HAPPEN to coincide with
yours, aren't as good as one big guy who genuinely cares about you.
or so says my idealist streak.
justin
_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:36 MDT