RE: flare and SIAI

From: Xavier Lumine (dmyshkin@hotmail.com)
Date: Sun Jul 29 2001 - 17:27:19 MDT


>My feeling is as follows:
>2) goal-driven, intelligence-amplifying self-modification involves so many
>hard AI tasks that the task of dealing with a tough programming language
>like C is basically trivial by comparison.
>In support of claim 2, I observe that, for humans, learning to write
>complex
>AI code well involves many, many things more difficult than mastering a
>programming language. I suspect it will be the same for a
>code-writing/modifying AI.
>

Yeah, true. However, it is my belief that if the AI project is going to
require a very large codebase, a language suited for the project is in
order. You could write a text-parser in java instead of Perl, or an
operating system in assembler instead of C, or a word-processor in LISP
instead of Visual Basic, but why waste programmer effort? Although many
complex AI coding concepts will have to be implemented, why make the
developers' work even harder than it already is (very, very hard).

>I'm hoping that Eli's AI project will reach the coding phase *well before*
>the Flare project is done. If this happens, then there will be a time when
>your effort will be going to Flare when it could have been going to Eli's
>AI
>system itself. Whether this sucks or not depends on how many others are
>working on programming his AI system, obviously. A handful of people on
>his
>prototype AI system is surely going to be enough for a while, so if there
>are more volunteers than that, it makes total sense to have some people
>working on side projects. I hope it comes out that way, and there are more
>than enough expert coders to go around!

I'm the project leader for Flare, which means that at some point there will
be enough developers (hopefully) working on the project that my
project-management functions won't be so overbearing that I won't be able to
help work on the AI when coding begins. Also, we're trying to push out an
early Flare while the AI is still in concept and planning (i.e. push out the
early Flare in six months or less), so the AI developers will be able to
work in primitive Flare instead of having to write Python code and then
translate anything productive into Flare as that becomes available.

_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT