From: Peter Voss (firstname.lastname@example.org)
Date: Wed Jun 13 2001 - 13:56:40 MDT
Sandboxed AIs are not really the issue. The most advanced AI's will be those
that freely interact with the Internet - that's how they best acquire
knowledge, form concepts, learn skills, etc. Any design that is not
connected, is badly handicapped. Creating complex virtual worlds is too
expensive, and in any case, they cannot properly simulate human psychology
(and lots of other things).
We will have to find other ways of predicting neutrality/ benevolence. My
guess is that assurances will come from philosophy/ epistemology/ ethics and
psychology - superimposed on computer science. They may provide the theory/
www.optimal.org - Any and all feedback welcome: email@example.com
From: firstname.lastname@example.org [mailto:email@example.com] On Behalf
Of Jimmy Wales
Subject: Re: Effective(?) AI Jail
But there another argument about why we ought to be concerned about how to
keep a superintelligence "in a box" for awhile.....
Eliezar runs from the box, grabs a shotgun, and puts this thing to sleep
fast. Whew. Time for a new design.....
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:36 MDT