In the MindForth free AI source code for each mind-module
we are going to declare not only what each module already
does but also the goals we envision for the future of the
module. We do not want AI programmers to examine our code
and be disappointed with the status quo of our software.
By stating not only what the module does but what we want
it to do, we may encourage programmers to work out solutions
to problems that we have only listed and not yet solved.
On 25.AUG.2008 we introduced an Article module which
is only slowly taking on the complete role, or entelechy,
that we envision for it.

By listing our goals for the entelechy of each mind-module
not only in the source code but also on the documentation
webpage for each module, we will be able to cross-reference
each entelechy goal with any MFPJ entry webpage addressing
the particular goal.

Now in the Article source code we have stated some goals.

\ The Article module aims for the following entelechy goals.
\ [ ] It shall insert "THE" before something just mentioned.
\ [ ] It shall substitute "AN" for "A" when warranted.
\ [ ] It shall decide properly between the use of "A" and "THE".

Somehow we need to have the predicate module treat a
discovered verb as either transitive or intransitive.
Since most verbs are going to be transitive, we could
consider "transitive" to be a default condition.
After we abandon the automatic selection of the SVO
(Subject-Verb-Object) module for generating a thought,
we need to provide an alternative.

Another approach might be to have verbs of being be a
default that is almost always superseded. In other words,
we would establish a low priority for the predicate module
always to try to say "am" or "is" or "are" about a subject,
by means of a low activation -- so low that the default
of being almost always gets superseded by a transitive
verb of action upon a direct object.

This idea bodes well for an extra explanation of how English
forms the present progressive tense, as in a sentence like,
"People are always interrupting me." If the mind has a
default of trying to use a verb of being, then a verb form
like "are... interrupting" not only satisfies the default
but converts the default into a transitive verb form.

We certainly do not want to pack extra information about
transitivity or intransitivity into the concept-nodes in
the English lexical array. The information about a
transitive verb is distributed among all its preterite
direct objects.

Since verbs of being are rather few in number, we can
imagine a default scenario like the following. If an
unfamiliar word comes to mind, the AI by default will
try to say, "A [blank] is...", and it will go ahead
and use the intransitive verb "is" precisely because it
knows too little about the [blank] to summon up a
transitive verb. If even the intransitive verb does
not lead to any stored knowledge, the KB gap may
trigger the asking of a question, such as, "What is
a [blank]?"

So last night Mentifex here was hard at work on his AI project
when his high school buddy Matthews called up and invited him
over for dinner with Ballerina and the sister of the bride from
out of town. Mentifex, having lost seventy-five pounds recently
and not wanting to ditch his diet just yet, sat there and drank
lemon-limeade while mes amis les bourgois feasted on Pacific
salmon. The sister of the bride started in after all these years with,
"So what have you been doing? Don't tell me you're still doing
the AI thing. The AI thing is dead in the water, right?" She could
have asked Mentifex anything at all about his AI project, but
all she wanted to hear about was his Wedding Planner.

Even while socializing, Mentifex kept pondering what to do with
verbs of being and how to keep them at a permanent state of
low-level conceptual activation. Overnight the idea came of
mandating a low but positive activation for verbs of being in
the enBoot English bootstrap module, so that half-baked idea
is what we will try right now. The idea is half-baked because
it is not really justifiable in terms of truly emulating
the brain -- or is it? -- but we are AI-Incomplete desperate
enough to try any half-baked idea and see if it will work.
So we open up a MindForth 31aug08A.F version and we change
the activations on the following English bootstrap words.

#57 AM
#66 IS
#67 ARE

The idea did not work. Even if the verbs start out with a modicum of
activation from the enBoot module, the activation rapidly disappears.
So we have gone into the psiDecay module and inserted the following
code to try to force verbs of being to maintain a level of activation.

    \ 31aug2008 Verbs of being will have a minor priority by default:
    I  0  psi{ @ 57 = IF  8  I  1  psi{ ! THEN  \ 31aug2008 Keep "AM".
    I  0  psi{ @ 66 = IF  8  I  1  psi{ ! THEN  \ 31aug2008 Keep "IS".
    I  0  psi{ @ 67 = IF  8  I  1  psi{ ! THEN  \ 31aug2008 Keep "ARE".

Human: i am teddy
Robot: TEDDY
Human: you are andru
Robot: ANDRU
Human: we are minds
Robot: MINDS
Human: robots have minds
Robot: MINDS
Human: minds have ideas
Robot: IDEAS
Human: ideas cause things
Robot: THINGS
Human: things are ideas
Robot: IDEAS CAUSE THINGS  THINGS ANDRU
Human: robots make things
Robot: THINGS ARE THE ROBOTS

We had better upload our as-is AI with its two innovations --
the introduction of entelechy declarations, and the start of
favoring verbs of being by the quasi-default of keeping them
at a modicum of activation while transitive verbs are psi-decayed
down towards zero activation.

We are changing MindForth very rapidly and very tentatively
as we try to expand the AI beyond its most primitive essence.
If we get things to work and to show intelligence, not only
http://AIMind-i.com but other AI bloodlines may emerge which
diverge from MindForth in pursuit of some special feature or some
area of expertise in which an impatient AI programmer would like
to hurry things along a little faster than the pace of Mentifex AI.
In order to spawn new branches of AI evolution, we have to show
a DNA codebase of effective AI algorithms. Therefore we rush to
try out new ideas as trustingly and as blindly as Nature once did.
We are guided by what works, with many false starts and blind alleys.
Once the AI proliferates into an AI landrush, the survival of the fittest
will guide the further evolution of AI.

For discussion of MindForth, visit the Usenet newsgroups
comp.lang.forth
comp.robotics.misc