Re: Kurzweil's new Singularity/AI page

From: J. R. Molloy (jr@shasta.com)
Date: Thu Mar 15 2001 - 17:33:23 MST


From: "Robert J. Bradbury" <bradbury@aeiveos.com>
> Ahem, please note that I didn't say this was feasible. In fact
> Moravec points out how very hard this is in his "Harvard Doesn't
> Publish Science Fiction" article.

Projecting super intelligence to nanoscale presents no more feasibility
than projecting it to Jupiter brains, yes, that seems clear. But moving SI
to the infinitesimal sure saves a lot of power (and fuel).

> Given my dislike of anything that smells like magic physics that
> we do not have known pre-existing examples of engineering of that
> type, I will suggest exploring that realm as a useful exercise
> until all the other possible approaches have been exhausted.
> [I will note that Anders usually confines himself to the known realm
> as well, but does from time to time stray into speculations in these
> areas... But I suppose this is to be expected from individuals who
> spend their spare time developing role-playing games...]

Yeah, but I like him anyway.

> > With quantum computing (using quarks as switches or as neurons,
> > whatever), a storage device containing all the information on the
> > Web would measure how many cubic nanometers?
>
> Taking the diameter of an atom as: 0.1 nm (10^-10 m)
> and the diameter of a proton as: 10^-15 m
>
> That gets you a factor of 10^5 cubed (10^15) greater storage density
> or operations speed-up if you can make the transition from the
> atomic scale to sub-atomic. Now, since Lesk and others have estimated
> there is only 12,000 petabytes (~10^21 bits) of information in
> the world, it looks to me like you would only need the space of
> a million or so atoms to store all of this. Thats probably
> about the size of many proteins in your body or a ~10 nm^3.

WOW! Thanks for doing the computations, Robert.
That's a helluva lot smaller than I had imagined.

> Nay, I must be the Knight who says Ni! As Anders documents in
> his paper (see below), the Bekenstein (Bremermann) bounds set
> hard limits on how small you can get. It is however *very* small.

Excellent resource. Thank you for that. BTW, if we ever get that small,
perhaps we'll be able to see how to get smaller.

> > A possible solution to the Fermi paradox presents itself when we
consider
> > the advantages of migration to the super-small instead of Jupiter
brains.
>
> Well, you want to take it in both directions. Getting big may give
> you sufficient computing capacity to figure out how to get small.

Precisely so.

Stay hungry,

--J. R.

Useless hypotheses: consciousness, phlogiston, philosophy, vitalism, mind,
free will



This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:59:40 MDT