RE: One or more FAIs??

From: Ben Goertzel (ben@goertzel.org)
Date: Thu Jun 03 2004 - 12:54:51 MDT


Of course, hardware vs. software presents a sort of chicken-and-egg
problem for AGI...

Until we understand AGI in more detail, we won't really know what the
hardware requirements for human-level AGI are. They could be a single
modern desktop PC, or 10^10 of them. My guess is a few hundred, but
it's not so solid a guess.

On the other hand, if we had a lot more and better hardware, we could
run a lot more tests of our AI approaches a lot faster, and we could use
programming languages better suited for rapid development. We use C++
for Novamente because it's so good at running fast while manipulating
massive amounts of memory, but if hardware were better we could use a
more rapid-dev language while still getting excellent performance.

So, having more hardware would let us move a bit faster toward
understanding what the hardware requirements really are!

-- Ben G

> -----Original Message-----
> From: owner-sl4@sl4.org [mailto:owner-sl4@sl4.org] On Behalf
> Of Metaqualia
> Sent: Thursday, June 03, 2004 2:35 PM
> To: sl4@sl4.org
> Subject: Re: One or more FAIs??
>
>
> > >It's not about the hardware. It was never about the hardware.
> > But it could be. I'm not really expecting it but it's possible
> > tomorrow it will be announced that a way has been
> discovered (perhaps
> > with non-Abelian particles) to make a practical quantum
> computer, and
> > then the world will never be the same.
>
> I agree.
> Your gameboy ain't going FOOM no matter what you program into
> it. And I would like to quote a contradicting statement by
> Elizier himself "Moore's law keeps lowering the bar"...
>
> mq
>



This archive was generated by hypermail 2.1.5 : Tue Feb 21 2006 - 04:22:38 MST