Re: analog computing

From: Gordon Worley (
Date: Sun May 13 2001 - 08:47:05 MDT

At 1:18 AM -0700 5/13/01, Aaron McBride wrote:
>want to know 100% what the result of some operation is/will be...
>But if the only difference between today's computers and tomorrow's
>is speed and storage space (I think this is what Gordon Worley was
>getting at) then why don't we just write the AI to bring about a
>Singularity now (or last week)? If it's all deterministic anyway,
>then couldn't we just simulate The AI with pencil and paper? What
>I'm getting at is there seems to be some ingredient missing...
>something that will make the computers of tomorrow be able to do
>more than just play a really really fast game of solitaire. Maybe
>all we're missing is the speed and storage space (and human
>invention in the way of software), but we shouldn't rule out the
>possibility that more will be required.

Yes, this is what I wrote. At the moment, though, the Singularity
couldn't happen that way. The reason being that the AI would be
slower than humans doing the same work. It might take 18 months for
humans to double the power of processors, but right now, if we got an
AI working that's sole function was to work on making better
processors (and being Friendly), on current hardware it would still
take too long. That's why there so much conern about programming AIs
more effeciently (see the recent threads about Webmind) and creating
more powerful computers. I won't deny that a move to quantum
computing or some other processor model *could* offer a jump in power
by a few hundred orders of magnitude, but running on that hardware
wouldn't be any different than running on the chips we have now, just
at the same order of magnitude (well, more or less).

I know that this is far below SL4, but I'll post it for the newbies'
sakes. What you are writing about is AI mysticism, the idea that
there is some kind of magical ingredient you have to put in the soup
to get an AI. As much as the media, society, SF authors, et al. have
spread this meme, there is no good reason to believe it to be true.
While it's romantic to think that there is some magical piece to the
puzzle of intelligence, and when it's found we'll get this amazing
hard takeoff, the truth is that, if such a thing even exists, it's
probably a line or two of code!

But, if your believing in AI mysticism gets me a quantum computer
sooner, then, by all means, ignore the previous paragraph. ;^)

Gordon Worley
PGP Fingerprint:  C462 FA84 B811 3501 9010  20D2 6EF3 77F7 BBD3 B003

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:36 MDT