From: Ben Goertzel (ben@goertzel.org)
Date: Tue Nov 01 2005 - 12:58:32 MST
Hi,
> As for 'emergence' from the Internet, speculation at the level of
> fluffy generalities is absolutely useless. If you think there's a
> specific mechanism that could conceivably produce an AGI distributed
> over the Internet, describe it, and we might have something to
> discuss.
>
> * Michael Wilson
I discussed this possibility in my 2002 book "Creating Internet
Intelligence."
I do think it's a real possibility, but, certainly not in the next
few years.
How it could happen would be if a bunch of narrow AI programs were
posted on the Net and allowed to interact as "autonomous agents."
Potentially, their interactions with each other and with humans
could cause the emergence of an additional layer of intelligence
not envisioned by the designers of any of the narrow-AI agents
involved.
I could spell out some specific scenarios via which this could
potentially happen. Probably others could too. You need to
envision agents using powerful narrow-AI tech that can do things
like
-- automatically summarize documents on the Net and post its
summaries on the Net
-- some simple computational reasoning based on information
extracted from documents and databases online
-- make economic and financial predictions and post them
online
-- do economic transactions such as financial trading or
automated product-purchasing, based at least partly on
information gathered from the Net (much of which was
posted by other AI agents)
-- give people recommendations regarding what to read,
what to buy, etc. with much more savvy than current
software
-- make scientific hypotheses via integrating knowledge
from existing databases (there is loads of scientific
data online)
I'm not talking about the emergence of intelligence
from unintelligent software like Word and Firefox
and Mathematica. I'm talking about the emergence of
autonomous intelligence from autonomous narrowly-intelligent
software components that interact tightly via the medium
of the Net. Potentially, in this sort of scenario, one could
see the self-organization of a kind of emergent meta-mind.
How likely is something like this? Well that depends upon
a lot of things, including the specific nature of the narrow-AI
software we're talking about.
My own bet is that someone will intentionally create an AGI
well before we get the chance to see whether this kind of
semi-spontaneous crystallization of AGI from the Net really
occurs.
-- Ben G
However,
This archive was generated by hypermail 2.1.5 : Tue Feb 21 2006 - 04:23:18 MST