Diagramming Robust AI

mentifex at scn.org mentifex at scn.org
Mon Mar 23 21:08:30 EST 1998

In friendly response to Mark Watson, Seth Russell, and Sergio Navega,
there are several good reasons for  writing a "Mind" program in Forth.

>Mark Watson wrote:
>> If I remember correctly (I have not used Forth since about 1980),
>> MVP Forth is available on many platforms, but since you are clearly
>> spending LOTS of your time on this project, I would like to make
>> a suggestion:  Do not use Forth because:
>> 1. Forth is very difficult to read and understand (unless you
>> wrote the code -- in the last few days!)
      True -- but
<a href="http://www.scn.org/~mentifex/aisource.html">Mind.forth</a>
has extensive comments and documentation, with HTML hypertext links
embedded INTO the source code for extreme ease of on-Web perusal.
>> 2. Pick a more popular language (like Java, Prolog, Lisp, even
>> perhaps C, C++, or VisualBasic) so that more people can try
>> out your system.
      Although I (Arthur) know only REXX, Smalltalk and now Forth,
other persons may port the AI further if it seems worthwhile.

>> Anyway, pardon me for poking my nose into your project.
           Oh, no -- thank you for the advice.  - ATM
>I must agree with Mr. Watson. I think that, being a difficult
>problem as AI is, we would be better off making its programming as
>confortable to humans as possible, not to machines. So, choosing a
>language like C++ is a better way to go, because it's easier to
>share work and there are a great amount of available libraries,
 ^^^^^ ^^^^ but first we appeal to the fanaticism of Forth devotees!
>that simplify implementation.
>Sergio Navega.

The main reasons for expressing the Theory of Mind in Forth are:
Robotics!  (Forth is the language of choice among robot builders.)
MISC!  (Minimal Instruction Set Computing)  Inexpensive Forth chips
     are being designed and manufactured.
Modularity!  All Forth programming is structured programming.
Domino series:  First Mind.rexx, then Mind.forth, then Mind. ....

More information about the Neur-sci mailing list