Re: Why would AI want to be friendly?

From: CYMM (cymm@trinidad.net)
Date: Mon Oct 02 2000 - 06:28:47 MDT


EUGENE quoting CYMM wrote:

> But suppose the machine evolves ten billion times faster - suppose the
> mutation rate is ten billion times that of a human - the boundary
conditions
> might preclude the darwinian selection that allows for concomitant
coupling
> of the two species' evolution.

EUGENE SAID: Huh? Both systems still evolve. (Of course, you have to have a
population, not a single individuum for it to happen). They still
couple, by virtue of occupying the same timespace. Of course, because
their fitness is not equal, one of the players is going to bite the
dust.

CYMM SAID: Eugene, I was commenting on the likelihood, a la Barbara, that
there could be "cooperative co-evolution". If we create an appropriate
"response function" for each evolving species; then we'll have to have some
coherence for non game theoretic cooperation to occur... this is so, BTW,
even if the response function is a quantum state vector - it's the case in
ferromagnetism; BCS superconductivity etc.

I do agree with you, of course, that coevolution will always occur once two
species share the same environment. But Barbara was talking cooperative
co-evolution.

EUGENE SAID: "...Radiating species attempt to fill all available niches.
Even if one
class of niches is way cushier, they're expanding as a random cloud in
the genotype/phenotype space, and hence will eventually fill them all.

CYMM SAYS: The keyword is "available". If a peak in the fitness landscape is
already occupied... it often does not pay to try to take it (...in a
radiation scenario with depauperate biota..) there are so many other peaks
nearby. It is only when the holder of the peak is (comparatively) very pooly
adapted would a conquest occur. It's an annealing scenario.

Look at freshwater cichlids (...fishes...) in Central America... they
certainly had one heckuva time in the freshwater environment...but their
evolutionary forays into the mangrove estuarine & reef environments were
nowhere as successful - because of the presence of centropomids and
lutjanids in the estuaries... and the pomacentrids on the reefs. But there
was nothing else that stopped them... they were physiologically preadapted
to take such niches (... a few species have made tentative forays into such
niches - but it always pays to differentiate away...).

What I sincerely hope is that when the inevitable foray into exponential or
hyperbolic expansion of "intelligence" occurs - that the available
technology allows a lightning fast evolution... Methinks that a relatively
slow and protacted evolution in the early stages of such AI will put human
civilization at serious risk.



This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:31:20 MST