From: Justin Corwin (thesweetestdream@hotmail.com)
Date: Sun Apr 07 2002 - 17:41:37 MDT
>From: Dani Eder <danielravennest@yahoo.com>
>An AI may be able to convince it's programmers to
>buy hardware to upgrade itself, or it may do useful
>work to pay for it's upgrades. Then again, it may
>have to build up an industrial infrastructure to
>build parts no one else makes yet. Or it may take
>a lot of computer hardware to run an AI. In the
>latter cases a replicating factory with a short
>doubling time would be useful.
Are there any hard figures on what kind of matter/energy consumption a
general assembly macrofactory would entail? My gut instinct is that a
factory designed to build anything (within bounds) would be more
inefficient, due to the large design space.
Also, bounds are neccesary in order to fulfill the "Santa's Workshop"
hypothesis. Must it be able to make .13 micron computer chips? FPGAs? What
about ocean-going ships? Or just children's toys? ;-)
>
>So until someone shows that an AI operating in the
>realm of thought alone is sufficient to implement
>a singularity, I consider the hardware issues to
>be relevant.
>Daniel
I personally believe that a FPGA workstation with a decent amount of memory
would be dangerous enough, but I would expect that the SI would rather have
a STM than a macro factory, if push came to shove.
I'm more interested in Santa's Workshop for myself, actually. Something like
that would be very useful to have around. In fact, a suitably efficient
universal factory would probably be sufficient to support a small state.
(visions of sugarplums...)
Justin Corwin
outlawpoet@hell.com
_________________________________________________________________
Send and receive Hotmail on your mobile device: http://mobile.msn.com
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:38 MDT