From: Bill Hibbard (test@demedici.ssec.wisc.edu)
Date: Sat May 15 2004 - 12:04:53 MDT
Hi Philip,
> Anyway I guess my final feeling thoght is that your method could work
> while AIs and humans are confined to the Earth but it is likely to break
> down once this is not the case - especially if the AIs proliferate through
> the universe. How would you see you method working in such a
> 'geographically' non-constrained situation?
One possibility is analogously to the way it works in
humans. We value the happiness of other humans (for
example, mothers value the happiness of their children)
and we learn to connect the identity of those other
humans with their changing appearance and to remote
communication with them (such as via telephone).
Furthermore, we internalize the values of their
happiness. For example, I had a memorial for my mother
at her church, despite the fact that there were no
other surviving family members and certainly a church
service means nothing to me, just because it was what
she would have wanted. I think we can design AIs to
learn to identify uploaded humans as humans whose
happiness they should value, and uploaded humans may
travel in ways that are impossible for natural humans.
Cheers,
Bill
This archive was generated by hypermail 2.1.5 : Tue Feb 21 2006 - 04:22:36 MST