Re: singularity logic loop

From: Samantha Atkins (samantha@objectent.com)
Date: Thu Apr 25 2002 - 00:08:20 MDT


Lee Corbin wrote:

>
> There are just two reasons that it might be nice to
> humans, so far as I know: one, I. J. Good's meta-
> golden rule---be nice to sentients far below you
> who're at your mercy so that your rapidly evolving
> descendent (who soon won't identify with you) will
> be nice to you---, and two, because someone built
> it to be that nice.
>

Or it decides that non-arbitrary ethics preclude destroying
other sentients as much as possible.

> You're right about economic pressure, though. As the
> speed of light becomes so slow (from the SI perspective)
> evolution even over the Earth could break apart and become
> localized quickly, and in that case systems that allowed
> anything to hold back their own development would be at
> a serious disadvantage.
>
>

This has the built-in assumption that endless-growth is such a
strong and mandatory drive that all else is subservient. I
don't think this is a given. A serious disadvantage relative to
what? There are no other SIs in the neighborhood we can detect
and it is not a given that they could think of nothing better
than endlessly competing with one anothr throughout space-time.

> Yeah, but only insofar as it being kind of funny (for
> me any way) to imagine an AI that took over but didn't
> want to mess with any of the Earth's surface because of
> humanitarian or environmental concerns ;-)
>

I don't see much humor in presupposing the opposite personally.

 
>
> ;-) I was thinking that they'd leave us the top 0.1% as
> part of the gag. That would be several miles deep. But
> seriously, of course making a Dyson sphere (or sheaves, or
> whatever Robert calls them) would be a priority for our
> hungry entity.
>

Precisely why should it be that hungry or that hungry that
quickly? Cancerous levels of growth until all resources are
consumed are simply not the only viable models of Singularity
level beings.

- samantha



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:13:39 MST