From: Robin Lee Powell (rlpowell@digitalkingdom.org)
Date: Wed Aug 22 2007 - 12:34:22 MDT
On Wed, Aug 22, 2007 at 11:22:29AM -0700, Eliezer S. Yudkowsky
wrote:
> Rolf, you want to re-explain your proposal? I think a lot of
> people didn't get it the first time.
I'm pretty sure I got it, I just wonder if an AI so broken as to
engage in subgoal stomp would be dettered by a threat to the
completion of that subgoal. IOW, given the former I assume the AI
is really pretty stupid in important ways.
Having said that, every detterent we can get their hands on should,
of course be used.
-Robin
-- http://www.digitalkingdom.org/~rlpowell/ *** http://www.lojban.org/ Reason #237 To Learn Lojban: "Homonyms: Their Grate!" Proud Supporter of the Singularity Institute - http://intelligence.org/
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:58 MDT