Any progress towards AI at all? was Re: [extropy-chat] Futures Past

Dirk Bruere dirk.bruere at gmail.com
Mon Oct 10 01:14:37 UTC 2005


On 10/10/05, J. Andrew Rogers <andrew at ceruleansystems.com> wrote:
>
> On 10/9/05 5:46 PM, "Dirk Bruere" <dirk.bruere at gmail.com> wrote:
> > Well, 10 PFLOP machines are on the boards for implementation by 2011.
> > That, IMO, is Human Level processing power.
>
>
> If we are going to use such a lazy and ambiguous definition of "machine",
> we've had 10 petaflop machines for quite a while already.
>
> And asserting that any such thing is "human-level processing power"
> suggests
> that you have far more information than you do. If you could make that
> assertion authoritatively, you could also produce an AI to demonstrate the
> point.
>

Since I don't have the required s/w obviously I can't.
We are talkinmg about h/w processing power - remember?
Various microcephalic cases with normal functioning indicate that only some
10% of the brain mass is actually required for a fully functioning Human
(references posted some time back in a previous discussion). I *suspect*
that if only the intellect were required that could be reduced by another
factor of two or three.

10^10 neurons, 10^4 axons, 10^3 Hz etc
10 PFLOP is in the right ballpart.
And if it isn't we'll just have to wait 10yrs for another factor of 100
increase.
That's a general type of processor array. No doubt with dedicated h/w we can
multiply it by at least another order or two of magnitude.
That's still only 14yrs away.

Dirk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20051010/928fa5e9/attachment.html>


More information about the extropy-chat mailing list