1.
Theory of
AI Textbook Algorithm Steps: Code the Think Mind-Module
/^^^^^^^^^^^\ SVO = 1:Subject +2:Verb +3:Object /^^^^^^^^^^^\ /visual memory\ | | | __________ / auditory \ | _______ | | | | / Sentence \ | memory | | /"cats" \ | | | | \__________/-------|-------------\ | | / percept \---|---|-+-|--------/--\-----------|---c-a-t-s | | | \ engram / | e|c|f| ____/ \____ | | | | \_______/ | a|a|i| /Noun\ /Verb\ | | | | | t|t|s| (Phrase) (Phrase) | | | | | |s|h| \____/ \____/ | | | | | | | | 1|Subj. / 2|Verb | | | | | | | | | Obj./3 | | | | | _______ | | | | | / _|___ | ________ | | | /fresh \ | |_|_| _|___/ / En \ | / \ | | | / image \ | / \ / En \ / Verbs \-|-/ "eat" \| | | \ engram /---|--\ Psi /-/ Nouns \ \_______/ | \ "fish" / | | \_______/ | \___/ \_______/------------|--\________/ |
The diagram above shows how syntax flushes out nouns and verbs to
generate thoughts. The high-level Think module calls lower-level
modules such as English, the subject-verb-object SVO module,
the Noun-Phrase module and the Verb-Phrase module.
The auditory memory channel is where the Mind hears itself think,
because the mechanisms (modules) of thought operate in concert to
manipulate auditory memory engrams and string words together into
a sentence that may or may not be spoken by the voice of speech.
As each thought is generated, it leaves swirls of activation in
the abstract Psi concepts that either found expression in thought
or were logically close enough to the generated thought to be
partially activated and thus be ready to figure in a succeeding
thought as the moving finger writes and, having writ, moves on.
2. I Think, Therefore I Am
There is no consciousness module in the artificial Mind, because
consciousness is an emergent property emerging from the nature of
every Mind as a phenomenon greater than the sum of all its parts.
Consciousness is therefore an epiphenomenon of the thinking Mind.
But what is thinking? In artificial intelligence, the activation
that spreads from concept to concept in a chain of associations
may often be expressed either as a series of remembered images in
the imagistic reasoning of visual thinking, or as verbal thought
in the form of a sentence generated by a syntactic superstructure
that flushes out all the most highly activated concepts, reifies
them into nouns and verbs and other parts speech, and speaks them
in communication with other thinking minds or in personal silence.
By thinking, the creature knows that it must exist, if it thinks.
3. Quot Homines, Tot Sententiae
However many humans (there are), so many opinions (there are).
Luckily, human minds do not think and speak in their own ideolect.
Instead, many minds share a common language, and at worst there
are dialects that differ from one region of a language to another.
Possibly all human languages arose from some faintly discernible
Ursprache or proto-language, but we futurists disdain and dischew
the lifelong quest of ungravitated linguists who eschew AI Minds
in favor of the unfathomable past because they have not properly
gravitated to the grandest of all intellectual Grand Challenges:
the passing of the torch from human to machine intelligence in AI.
Let the dead bury the dead, and let us the living create AI Alife,
where first among our concerns is how to engender thinking in a
machine, and where the first function of the Think module is to
decide what language to think in.
Although you may program the artificial Mind to think and speak a
particular human language, you may go beyond one native language
and you may easily impart to your intelligent robot the ability
to think in several natural languages. The organization of the
AI in accordance with a linguistic theory of mind (TOM) makes the
robotic psyche a natural instrument for machine translation (MT).
Since there must be only one deep conceptual mindcore but there
may be multiple lexicons and multiple syntactic superstructures
in the AI Mind, it may think the same thoughts in any language
that it learns from you or from its World Wide Web environment.
You only need to code the appropriate triggers to select whatever
language the machine is ready to respond in by dint of containing
a full complement of lexical bootstraps and syntactic structures.
Normally an AI will start thinking and communicating in a default
language until it perhaps encounters foreign-language input which
is recognized and which brings a foreign-language apparatus into
play in the artificial Mind that continues to associate concepts
in the central mindcore -- independent of any particular language.
Another possible trigger for thinking in a non-default language
might be the contemplation of a subject matter or of a historical
figure commonly associated with a particular human language which
happens to be the main repository of all primary documentation or
resource data for the subject in mind. An AI Supermind will grow
so fluent in all past, present and future human languages that it
will simply not care which particular language you address it in --
as long as you yourself -- the hapless intruder upon the serenity
of the Semi-Omniscient -- are able to speak the intended language.
5. The Language of Pure Thought
If you have ever heard Sphaerenklaenge, the Music of the Spheres,
then you know that there may be spheres of expression beyond our
human ken and beyond our crude ability to organize reality into
woefully inadequate and confusingly ambiguous human concept maps.
Eye has not seen and ear has not heard the ideal language of Mind.
http://mind.sourceforge.net/ai4u_157.html
is an overview of Mind.
// Think() is called when the main aLife() function is in
// a state of "quiet" after cessation of any current input.
function Think() { // ATM 12aug2002; or your ID & date.
pov = "#"; // All thinking has "internal" point-of-view.
output = ""; // Attempt at persistent display.
act = 10; // or whatever threshold is to be tested for.
for (i=t; i>midway; --i) {
Psi[i].psiExam(); // Examine recent Psi nodes.
if (psi4 == 8) { // Only look at verbs (i.e., negatable).
if (psi0 == 59); // alert('Think: verb is DO'); // Test
else { // Look for main verbs; avoid "do".
if (psi1 > act) { // A check for possible chain reaction.
act = psi1; // Let the found value be the new measure.
psi = psi0; // Seize the psi # for Activate().
jux = psi2; // Checking for negative "jux",
// to cause the calling of negSVO syntax.
} // End of if-clause looking for high activations.
} // End of if-clause avoiding forms of "do".
} // End of if-clause looking for possible negative verbs.
} // End of backwards loop seeking high activations.
if (psi > 0) {
Activate(); // To reactivate the chosen concept.
English(); // Calling one or more choices of English syntax.
}
psiDecay(); // After thinking, let stray activations decay.
} // End of Think(); return to aLife().
\ THINK is called by the main ALIFE function.
: THINK \ ATM 29jul2002; or your ID & date.
35 pov ! \ All thinking has "internal" point-of-view.
10 act ! \ or whatever threshold is to be tested for.
midway @ t @ DO \ Examine recent Psi nodes.
I 4 psi{ @ 8 = IF \ Only look at negatable verbs.
I 0 psi{ @ 59 = IF \ If found verb is form of "do"
I 1 psi{ @ 10 > IF \ If activation is higher than 10,
10 I 1 psi{ ! \ 29jul2002 Test: reduce activation.
THEN \ 29jul2002 Test
ELSE \ Skip "do" and act only upon main verbs:
I 1 psi{ @ act @ > IF \ If activation found high,
I 1 psi{ @ act ! \ Let found value be new measure.
I 0 psi{ @ psi ! \ Fetch the psi # for Activate.
I 2 psi{ @ jux ! \ jux=12=not will call negSVO.
THEN \ End of test for high-activation verbs.
THEN \ End of if-clause avoiding forms of "do".
THEN \ End of test for negatable opt=8 verbs.
-1 +LOOP \ End of backwards loop seeking high activations.
psi @ 0 > IF \ If an active main verb is found,
ACTIVATE \ to reactivate the selected Psi concept.
ENGLISH \ to select an English syntax structure.
THEN \ End of test for a found main verb.
\ psiDecay \ After thinking, let stray activations decay.
; \ End of THINK; return to the main ALIFE loop.
8. Analysis of the Modus Cogitandi
The Think module thinks by expressing abstract concepts and
their relationships in natural, traditionally human but now also
robotic language. In the early stages of the AI, the thinking
is simple. As more and more modules for negation and questions
and additional Chomskyan grammar transformations are included,
the ability of the AI to think will become much more obvious,
but even the simple AI Mind thinks and remembers its thoughts.
Although the Think module may seem rather simple if it only calls
the English subroutine for verbal cognition, it is nevertheless
important to establish Think as a juncture or waystation from
whence several things may branch out in the future, such as the
choice of various human languages, or even a decision whether to
think in language or in visual imagery, musical tones, etc.
9. Troubleshooting and Robotic Psychosurgery
Try in advance not to introduce any
evolutionary bugs.
The
AI Debugger program may shed some light in general on how to debug
and troubleshoot programs in artificial intelligence.
10. Think Resources for
Seed AI Germination and Evolution
