Re: How to make a slave (was: Building a friendly AI)

From: Stathis Papaioannou (stathisp@gmail.com)
Date: Wed Nov 21 2007 - 00:05:25 MST


On 21/11/2007, John K Clark <johnkclark@fastmail.fm> wrote:

> Well, you seem to have goals, they must have come from someplace, I'll
> bet the AI's goals will come from the same place. Or are you one of
> those sacred atoms people who think meat is sacred and silicon profane?
>
> > and on what basis are the good goals
> > distinguishable from the bad?
>
> I imagine the AI will accomplish that task the same way you do, by using
> his judgment.

Human goals, values etc. have evolved to be what they are. You could
argue that when the world is thick with AI's of every stripe, those
with goals and values more conducive to survival and growth will lord
it over the AI's crippled by a willingness to submit to humans. But
this is quite different to arguing that simply by virtue of being
intelligent the AI will have particular goals and value. There is no
logical contradiction in having godlike powers and wishing to use
those powers in the service of less capable beings.

-- 
Stathis Papaioannou


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:00 MDT