From: Robin Lee Powell (rlpowell@digitalkingdom.org)
Date: Thu Feb 17 2005 - 13:54:02 MST
Someone just showed up on the Lojban list talking about using Lojban
with computers, including that he wanted to create a self-improving
neural-net AI.
I mailed him privately to suggest that neural nets were A Bad Idea,
and gave him some things to read from My Most Holy Lord Eliezer.
Excuse me, I need to surgically de-graft my tongue from my cheek.
There. OK, moving on.
He responded with "Well, it's going to take a long time for anything
interesting to happen", followed immediately by "Hmm, we're within
2000 or so P4 machines of brain parity; that's intense" (he got the
math *way* wrong). Then "Besides, AIs don't start with motivation,
so it's not like they'd want to kill us" (good, you're on the path)
followed immediately by "So I'll just make a green button that I'll
push when it does something good" (*ouch*).
Not making this up.
Not that it's at all unreasonable that someone would think of these
things, but just that it's *so* stereotypical a conversation. :-)
-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 : Tue Feb 21 2006 - 04:22:53 MST