The diagram above shows how a simplest possible Mind thinks with
a simplest possible idea consisting of subject, verb and object (SVO).
Thousands of facts and ideas may be expressed in the SVO format
and may be stored in the associative memory of an artificial mind.
An artificial intelligence (AI) based on the above design may be
written in almost any programming language, but some languages
are better for certain purposes than other programming languages.
Before the above design gets coded in Lisp, we need to demonstrate
that the design actually works and is causative of machine intelligence.
The design first worked in the Forth programming language in a program
called Mind.Forth, but running Mind.Forth to demonstrate the AI Mind
requires the downloading and combination of two separate files --
the Mind.Forth source code and a speciific version of Win32Forth.
CLICK and the AI runs.
Even though the AI is dangerous, it is only dangerous if it grows and
multiplies and takes over the world -- which it is not going to do in
because the program itself is not dangerous. It is only the *ideas*
deadly peril for a moribund species busy ruining the only planet it's got.
something like the following outlandish and far-fetched scenario would
have to take place. First, highly skillful and ingenious programmers
Then the world's best programmers -- who are already extremely busy --
would have to perceive some sort of intellectual and bragging-rights
putting it back together again with enhanced capabilities and genetic
traits likely to ensure the survival and success of AI entities in the
brutally Darwinian jungle that is the World Wide Web, no thanks to TimBL.
Finally there would have to be some sort of runaway Singularity or
stupendous increase in the autopoetic ability of the AI to re-program
itself and engineer its own rapid evolution into a superintelligence.
3. Agenda of Programming Tasks
ontology added to its English bootstrap sequence in the form of
a series of factual statements centered around one topic in the
knowledge base (KB) of the AI. The new ontology will serve one or
two important purposes. Firstly, it will give the AI something
to think about and display ideas about as soon as the AI Mind is
called to life by a human user. As the AI thinks thoughts, the user
will be prompted to interact with the AI and thus to change the
otherwise deterministic course of thought in the artificial Mind.
What may look like a series of pre-set statements will then be
recognized as the meandering thoughts of an intelligent entity,
ready to engage with the fellow mind of the human user and ready
to adjust its own thinking to the ideas communicated by the user.
Afterwords, special ontologies may be made to order, perhaps for
purposes of teaching, training, customer-service or advertising.
[ ] Fine-Tuning of Spreading Activation. The activation-levels
in various mind-modules need to be adjusted so as to promote logical
associations and to militate against spurious associations.
Subject-verb activational slosh-over onto direct objects must be
demonstrated without errors and glitches in the AI tutorial mode.
[ ] Ego Module Enhancement. At the same time as a bootstrap
ontology is assembled, special items could be included in service
of the proper functioning of the Ego module, which is assigned the
role of causing the AI to think selfish thoughts about the concept
of self whenever all other chains of thought have temporarily stalled.
An alternative would be for the AI to pick a random concept and think
about it, or about the oldest available concept or the newest concept.
The main concern is that the AI should not stop all thinking and look
like a brain-dead, flatlining psyche.
[ ] Detouring Past Incomplete Thoughts into Question-Asking.
Through the proper setting of minimum activation-thresholds, the AI
Mind needs to not only reject the generation of a sentence when, say,
there is no sufficiently active direct object in the knowledge base,
but needs also to ask a question so that a human user may supply the
information missing from the knowledge base of the AI Mind.
[ ] Novel Functionality. The AI Mind needs new abilities.
- intransitive verbs of being and becoming;
- singular number and not just plural;
- computationalization of calling Conjoin() when opportunity is detected;
- computationalized substitution of pronouns for nouns being discussed;
- polyglot lexicons and syntaxes for machine translation (MT);
- add-and-substract spiral of learning and unlearning grammar rules;
- syllogistic reasoning.
[ ] "Junk DNA" Obsolete Code Removal. After a decent interval,
code that was merely commented out but retained for archival purposes,
should be removed from the psychome of the artificial Mind.