PSS Ben I loved reading your blog. Pls keep it up. If you ever have time,
let us know why, of the 3 different AGI approaches you entertained, you went
with Novamente instead of the Hebbian neural net (and the theorem proving
one)... us scruffies would like to know... is it just your mathematical
bias/background or something more fundamental?



The Hebbian neural net approach seemed like it would be dramatically more
computationally expensive, requiring a whole bundle of synapses to do what
we can do with a single Novamente link.  I.e., it's less natural for the von
Neumann infrastructure we are stuck with at the moment.  And, once you get
beyond simple stuff, we don't know how the brain works so we need to invent
stuff anyway, even in that plan (e.g. I have a scheme for doing higher-order
logic in neural nets that involves feeding a dimensionally-reduced version
of a neural net's connection matrix to the same network as an input vector
... but tuning that would take a lot of work, and there is no neuroscience
to guide such work, at this point...)

The theorem-proving approach I rejected largely because it seemed there
would be basically no way to get significant funding for any incremental
results.  You'd be totally at the behest of government funding
establishments.  Yeah, once your system has proved the Riemann Hypothesis,
you're golden ... but just making "the world's best theorem prover" isn't
really going to get you anything but a few graduate students....  You can
paid more -- and more critically, from more different sources -- for making
a virtual big-scary-monster in a video game, than for making an automated
prover of moderately difficult set-theory theorems.  I am frightened of
being reliant on a single funding source, as NSF is moody: some years it
likes theorem-proving, some years not...

Also of course in the theorem-proving approach one doesn't have human
developmental psych to go on, in guiding the teaching process.  One can map
it into the new domain, but that's harder....

PSSS :) Google is doing narrow AI, Semantic Web & NLP, IBM is doing
WebFountain (i.e. also semantic web) and autonomous computing. So neither
seem to be in AGI. Anyone knows what M$ is up to? They have hired quite a
few smart CS *and* psych people too...


M$ Research is pretty much a traditional "big company" research lab, and
their researchers publish extensively on their work.  Lots of nice stuff on
Bayes nets, computational linguistics and computer vision for example.

None of these firms has started hiring up AGI researchers.  When they do
I'll be the first to let you know ;-)

-- Ben

-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e

Reply via email to