Re: Threats to the Singularity.

From: Stephen Reed (
Date: Thu Jun 13 2002 - 14:37:59 MDT

Cyc, which falls very short of AGI, is 1.4 million lines of translated C
code and the result of an 18 year project addressing only classical
symbolic knowledge representation and deductive N-order reasoning.

The KB (knowledge base) content can be roughly considered as source code
if the system could use its knowledge to improve itself (and it cannot
yet). Cyc has about 1.6 million assertions.

So I believe that the creation of Seed AI is far beyond the capability of
a single individual, and that belief is my motivation for working at

As a thought exercise I have wondered that if one can imagine a time in
the future in which AGI is prevalent, then imagine a homework assignment
given to a high school student: "achieve an AGI from scratch over the

Suppose they have a full understanding of the principles by which an AGI
worked, what few characteristics are required from which all others can be
automatically obtained by experimental self-improvement and from
learning by reading?


On Thu, 13 Jun 2002, Aaron McBride wrote:

> At 10:47 AM 6/13/2002 -0500, you wrote:
> ><clip>
> >
> >#2. One of a vast number of midnight hackers who have been studying AI
> >design over the internet hack together the first real AI. The only way
> >for one of the AI development teams to beat the hacker is to do it before #1.
> > <clip>
> >Mike.
> >
> >
> I was under the impression that this is about as likely as someone being
> able to "hack together" a 747 in their back yard. Building Real AI is
> probably not simple (we'd probably have stumbled on it by now if it
> was). I doubt anyone will be hacking together a Real AI on their own
> anytime soon. Anyone have estimates of the complexity of a seed AI?
> -Aaron

Stephen L. Reed                  phone:  512.342.4036
Cycorp, Suite 100                  fax:  512.342.4040
3721 Executive Center Drive      email:
Austin, TX 78731                   web:
         download OpenCyc at

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