From: Simon McClenahan (SMcClenahan@ATTBI.com)
Date: Tue Dec 31 2002 - 20:51:57 MST
On Tue, 2002-12-31 at 18:07, Wei Dai wrote:
> Eliezer suggests in CFAI that a Friendly AI may decide to divide the solar
> system into six billion equal parts and give each person property rights
> over a six-billionth share.
I haven't read CFAI in a long time, but to me this doesn't make sense.
No-one owns anything. Ownership is a human, maybe even animal, concept
that is a strategy for the strong to get stronger and hence the weak to
get weaker, So long as everyone is playing the same ownership Game. A
hostile intelligence would use the ultimate divide and conquer strategy
to claim ownership of all. Why can't we just share the whole Universe,
and why should a "strong" being who has a higher tolerance for survival
with limited resources be given equal or more than a "weaker" being with
a lower tolerance?
Semantically speaking, if an AI is going to be friendly to humans, it
should provide access to an abundance of resources, with no
over-consumption. Then maybe humans would learn to be friendly to each
other as well.
To quote from the recent (terrible) /Spiderman/ movie, "With power comes
great responsibility." A >H AI (or sysop? Are we still using that term?)
that has >H power will have responsibility, and humans will be the
benficiaries. Ownership will only create conflict. I know it sounds very
spiritual and philisophical, but there really is no law of nature that
includes ownership as a requirement for perpetuating life. It has caused
mostly problems for humans so far. I seriously doubt that an advanced
society would value ownership so highly.
If I get my fair share of space, I'll share it with anyone who asks for
it and is friendly enough to do the same.
cheers,
Simon
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:41 MDT