Re: AI timeframes

From: J. Andrew Rogers (andrew@ceruleansystems.com)
Date: Sat Apr 10 2004 - 10:58:46 MDT


On Apr 9, 2004, at 9:31 PM, Thomas Buckner wrote:
> One can guess that the government (parts of it, in any
> case) will be pushing for some sort of militarized AGI
> which, by its very definition, is selectively
> unFriendly toward individuals, groups or nations that
> the government views as a threat; one may even
> conjecture 'black budget' mil-AGI programs already in
> progress.

The primary interest of such technologies to the DoD would be for
sifting and analyzing vast quantities of information. That and
semi-autonomous control systems e.g. vehicles that can run themselves
without getting stuck in a ditch. It would not take the form of
Hollywood-esque Evil Supercomputer.

With only one exception that I can think of (the NSA), the DoD
outsources its science and engineering talent, which cross-fertilizes
with the private sector. It would be difficult for them to build and
isolate a sufficiently expert team to build a black budget AGI.

It would probably require that a noticeable number of people in the
community "drop off the map" in a particular country. There pool of
promising theoretical candidates just isn't very large as far as I can
tell, particularly when limited to citizens and such.

j. andrew rogers



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:46 MDT