From: christophe delriviere (christophe.delriviere@skynet.be)
Date: Fri Oct 23 1998 - 16:28:48 MDT
> Moreover; I would add that the multiple beings have one HugE
> advantage
> over, however wise and intelligent and whatever you can think of as a
> realistic attribute, one and only one individual: the multiple points
> of
> view.
>
> Even when a point of view is false, when parts of it include
> contrary
> beliefs, lies, redundancies, it's far more richer than one unique set
> of
> beliefs. An example: when a society of persons hold a belief contrary
> to
> its goals or its happiness, it's better than one individual set in the
>
> same kind of beliefs.
>
> Manu.
i'm not quite sure of that, an intelligent system with only one
identity, some part of this system who provide the illusion of being
one, probably to give to this system a quality of "adaptation" to his
environnement, can handle a lot of subagents, with something acting like
a central planner.
these subagents can compete and give their points of view. according to
certain rules (like darwinian evolution) the best points of view are
selected.
of course these agents can be very complicated and dynamically
evolutive.
i'm wondering what's the best topologies of intelligence could be, in
our solar system and his close surrounding, according to our actual
scientific worldview.
probably strong aglomeration of matter, to a certain extent provide huge
adventages.
but there is a compromise to reach between locality and dispersion.
we could imagine moon size computers surrounded by some kind of utility
fog build with something like the liliputian AI you were talking of.
or the moon size computers or asteroids used by a civilisation of small,
somewhat equals in their capacities, AI, as mere computers, tools.
with no embeded identity is the moon size, asteroids, or jupiter like AI
in such case.
i think of it and assuming the whole system is evolutive, i don't think
there is a easy to find configuration that is somewhat stable. Probably
there is some topological attractors in this problem anyway..
i would like to find more ressources according to this question...
is there somebody who can help?
-- DELRIVIERE CHRISTOPHE
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:49:41 MST