From: Ben Goertzel (ben@goertzel.org)
Date: Thu Mar 01 2007 - 20:06:25 MST
>
> What I was attempting to say is that, for purposes of saying "How can
> I disprove/prove that which you have failed to formally specify?", I
> will accept a disproof of stable iron-maximizing AI as a disproof of
> stable Friendly AI, and a proof of stable iron-maximizing AI would be
> huge progress toward a proof of stable Friendly AI. In other words,
> for purposes of preliminary discussion, you do not get to say
> "Friendly is ill-defined, therefore blah blah blah" because until you
> can do whatever proof or disproof you were trying to do with
> Friendly=<Integral IronAtomCount(t) dt>, there's no way you can do it
> for Friendly=<friendly>.
>
I understand your point, and it's certainly a sensible one, but I'm not
entirely sure it's correct.
Given a finite amount of computational resources, and a particular sort
of initial environment (e.g. Earth, human culture, etc.), it may be that
some particular goals are more easily stably achievable by AI's than others.
This is particularly likely to be true if one's AI works toward stable
achievement of its goal G partly via getting educated by humans.
Friendliness is more likely than iron maximization to get taught to an
AI mind via appropriate interaction with humans.
-- Ben
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:57 MDT