RE: Hardware Progress: $319/GF

From: Ben Goertzel (ben@goertzel.org)
Date: Fri May 31 2002 - 12:47:54 MDT


> Note that Human Equivalent Computing is only a comparison between
> benchmarked computer performance and theorized human brain computational
> power - not any measure of Artificial General Intelligence (AGI).
>
> -Steve

This is the key point. My guess is that we'll have adequate hardware for
human-level AGI *way before* before the mainstream AI community (a category
into which Cyc, in my view, squarely falls) comes anywhere close to a
plausible design for AGI.

The projects that have $15M for hardware in 2006, may be precisely NOT the
ones with any viable AGI design ;> Because AGI has been so unpopular in the
last 2 decades or so, that there is almost a contradiction in terms between
"big bucks" and "AGI focus." Although Webmind Inc. put a fair bit of $$
into AGI, our expenditure on hardware all total was under $500K, and we
couldn't afford much more, certainly not $15M.

In fact, I am guessing that Cycorp -- which has succeeded financially so far
in part because of its *non-AGI focus* (at least in its public face) --
doesn't have this kind of bucks for hardware at the moment either, though
they may have the connections to get it from the gov't with a sufficiently
good excuse and sufficiently avid politicking.

On the other hand, perhaps Novamente or A2I2 or Cyc or some other project
will attract enough attention via successes on modest computing platforms,
to convince corporate or gov't fund-meisters to invest in hardware for AGI.

Anyway, it seems clear to *me* that soon enough, inadequate hardware will
not be an acceptable excuse for the lack of viable AGI systems.... But of
course this is not proved, because these estimates of human-equivalent
computing are all guesstimates. Maybe the computing in the brain really IS
done in microtubules, we don't really know...

-- Ben



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:39 MDT