Re: AI Mind - FORTH

On Apr 22, 12:31 pm, Frank <fjru...@xxxxxxxxx> wrote:
Arthur, (ATM)

By accident I found a novel thing occuring with my AI Mind.
Remember my input is all buffered before processing and I added the
ability to execute Forth words (subroutines) from the user input
line. This lets me perform diagnostics while the AIMind is running.
Well, I was able to enter the following as a test:

Human: . : test1 cls cr 10 0 do i . loop cr ;

The '.' is a flag to process the line as executable. Well no errors
occured when I entered it.
Next I did:
Human: . test1 (cr) and the code was executed.

So what? Well to me it means the AI-Mind will be able to generate and
execute its own code, for that matter rewrite existing modules as
necessary. I just went in and reentered the test1:
Human: . : test1 cls ;
I was notified that test1 was redefined and when I called on it:
Human . test1 (cr)
and it did just as expected.

ATM (responding):
http://mind.sourceforge.net/forth.html is where I display what I
thought was necessary to achieve self-modifying code in Forth,
namely a looping spiral (for learning syntax) in which each
waystation on the loop is a node with about nine (9)
part-of-speech choices, each with a relative weighting,
so that a baby AI, like a human baby, could try out
various syntax sequences and gradually learn the ones
that meet with approval from the adult speakers.
So the code itself is not modified, but the transit
of the flow of control through the code is modified.
What you (FJR) have discovered. however, is news
to me. Would the wider Forth community please tell
us if Forth code can be modified on the fly?
And. if so, could not some file-writing routine
save the new Forth-words "to disk" so as to be
integrated into the AI Mind program as a whole.?


All of this is occuring while the AIMind is running. I have not
stopped the program and edited it and restarted it.

But If I do stop the program and do a warm boot, where it reloads the
mind values (CNS) I will lose the new definitions. Need to figure out
how I keep theses.

Still have an issue with it repeating subj and object being the same
noun. I am carefully going thru my code with both your java script
and your new version of forth side by side to see where or what I am
missing.
ATM:
Starting on May 5th I have a week to do a lot of Forth coding.
My ambitions for the
http://mind.sourceforge.net/mind4th.html (Mind.Forth)
program include seeking the range of working activation-
values for each setting in the AI so that I can then
use whatever seems to be the ideal (midrange) setting.
Then I want to remove any ad-hoc code that I kludged in
in order to get the Forth AI working as well as the
http://mind.sourceforge.net/Mind.html JavaScript AI.

Then I want to use the "pre" tags to make the AI
associate backwards from any entered noun
to any related verb, and further to any related subject,
so that the AI may think a thought using any tidbit
of knowledge available in its knowledge base.


Since my parser is able to identify adjectives, pronouns, adverbs,
conjunctions..... I need to do something in parser so 'pre - seq' get
connected to the subj & obj nouns and the verb and not point to
anything else in between.
ATM:
The original Mind.Forth parser is admittedly crude. I hope that
Forth AI gambits "take off" and stimulate coders to write
better parsers in the course of Forth AI evolution.


I only get to read my personal mail on the weekends at Yahoo. You can
use my work address if you still have it. I will be looking here and
on the win32forth group post.

Frank

http://tech.groups.yahoo.com/group/win32forth/message/12463
is what I saw first, surfing on in here from
http://mentifex.virtualentity.com/surf.html -- my AI surf page.
You go into more detail here, so I respond here.

http://mentifex.virtualentity.com/helpnews.html is a major
new link in my efforts to get people to pick up Forth AI
and run with it. On SourceForge I intend to post a series
of AI coding recruitment blurbs, each time announcing
whatever is new in the Mentifex AI project.

Right now the (publicly readable) Site Meter logs reveal
that many people are looking into both
http://mind.sourceforge.net/mind4th.html and
http://mind.sourceforge.net/Mind.html but
both the programs are still very primitive.

http://aimind-i.com (your AI Mind) has such
advanced features that I am very eager for
it to succeed and spawn many Forth AI descendants.

Bye for now,

Arthur T. Murray



.

Relevant Pages

  • AI Mind - FORTH
    ... By accident I found a novel thing occuring with my AI Mind. ... execute its own code, for that matter rewrite existing modules as ... I just went in and reentered the test1: ...
    (comp.lang.forth)
  • Re: OT: The Huyton axe murder made CNNs Headline News in America...
    ... Yes it would be a risk to execute people like the birmingham lads who were ... "A lot of football success is in the mind. ... had the best two teams on Merseyside, Liverpool and Liverpool Reserves." ...
    (uk.sport.football.clubs.liverpool)
  • Re: Parallelism in comparison sentences???
    ... I have long had the habit of using something for a ... I don't mind doing something the ... context if I've got to know the device first. ... What good is being an executive if you never get to execute anyone? ...
    (alt.usage.english)
  • Re: OT: Very Exciting News!
    ... NBC's programming is one solid 24 hour block of the Today/Tonight ... What good is being an executive if you never get to execute anyone?- Hide quoted text - ... I should have gone to bed before noticing the Kathie Lee thread again. ... Now I'll have to read something to get her perky self out of my mind. ...
    (alt.fan.letterman)
  • Re: Requirement ordering (was Agile developement ... more than just extreme programming ???)
    ... I hope you don't mind if I leave off the pretty symbology. ... a CFG system is not enough. ... Tomita context sensitive parser. ... Once you get the architecture right, ...
    (comp.object)