From: Eliezer S. Yudkowsky (firstname.lastname@example.org)
Date: Mon May 21 2001 - 19:05:55 MDT
Doug Keenan wrote:
> Hello everyone:
> OK, I RTFM. (Please note that "cacophony" has two O's.)
> Thank you Eliezer for making it available.
> There's no discussion of love or laughter here. Do we expect
> friendship to come before those, or not to see them at all?
Confuse ye not "Friendliness" and "friendliness". Um, basically, the
answer is as follows: If love or laughter are necessary to generate
Friendly behaviors, then the shaper/anchor semantics outlined in _Creating
Friendly AI_ constitute an unambiguous reference to them, and a Friendly
seed AI will perceive the need to rewrite itself to incorporate love and
laughter. If, as seems more likely, love and laughter are forms of
cognition unique to human altruists and are not necessary to altruists in
general, no such self-change would be made.
However, this doesn't rule out friendly, loving, laughing AIs. It just
means that such AIs are "citizen AIs" (or citizen superintelligences)
rather than Transition Guides or Sysops.
-- -- -- -- --
Eliezer S. Yudkowsky http://intelligence.org/
Research Fellow, Singularity Institute for Artificial Intelligence
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:36 MDT