From: Ben Goertzel (ben@webmind.com)
Date: Wed Aug 01 2001 - 06:10:27 MDT
> > And, so far as I've heard, no one on
> > here is building a "Codic Cortex" into their software.
> > I believe that is something that is expected to develop
> > eventually.
Actually we *do* have plans to build a codic cortex, but it only handles
"lower-level perception" of code; advanced cognition about code must be
handled by generic cognition methods that gradually get adapted to thinking
about code
The codic cortex embodies stuff like www.supercompilers.com are doing for
Java -- makes a nice labeled transition graph out of a program, i.e.
transforming software into mathematics that's easily digestible by an AI
system's reasoning engine. It does the same thing you need to do to prepare
a program for "traditional" global program optimization.
We need to do a lot of testing of general cognition on other problems before
hooking this up though. In fact, the new C++ Webmind doesn't even have all
cognition methods in it yet, only about 50% (the others are just a few
months off if things go well).
-- Ben G
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT