When the first true artificial intelligence,
MindForth by
Mentifex,
went operational in January of 2008 and started thinking after a
decade of arduous development, there was a companion program in
JavaScript called
Mind.html that ran directly off the Web in the
Microsoft Internet Explorer
(MSIE) browser. All a user with
MSIE
had to do was click on the link to see the
JavaScript artificial
intelligence
(JSAI)
flit across the 'Net and take up residence
in the Windows (tm) computer of the human user. It was so simple --
no programming involved, no set-up, no security worries, no need
of expert help -- like, for instance, a
docent at a museum.
But the JSAI tutorial program remains very limited in what it can do
and in what people can do with it. It is not suitable for installation
as the mind of a robot, because a JavaScript program is not allowed --
for security reasons -- to control anything but the Web browser on
its host computer. The JavaScript AI program also runs so slowly that
it tries user patience. The user waiting for a response from the JSAI
does not see the intensive computation going on behind the scenes as
the artificial
Mind races through its memory banks to think up a
response to an input from the user.
Nevertheless the
Mind.html JSAI is very good at what it is intended
to do. Since
JavaScript is a flashier, more visually appealing language
than staid old
Win32Forth, the JSAI serves its tutorial purpose
admirably. It shows graphically how an AI Mind thinks. It also
includes clickable links to other resources, such as the
User Manual,
the more difficult to install but intrinsically more powerful
MindForth,
and potentially to any science museum where users may visit
MindForth.
Thus the
Mind.html AI -- which is ridiculously easy to make copies of
and install on a Web site -- is out there on the Web, inviting users
to visit science museums in search of the real thing --
MindForth AI.
MindForth and its Mind.html JavaScript tutorial program are based
on a linguistic
theory of mind, unique to the Mentifex AI project.
Here is one sample brain-mind diagram from dozens of
diagrams
which explain the
theory of mind and which illustrate the
thirty-four chapters of the
AI4U book on artificial intelligence.
Mind.Forth and Mind.html think by activating concepts.
/^^^^^^^^^\ Hearing "c-a-t" Activates A Concept /^^^^^^^^^\
/ EYE \ _______ / EAR \
/ \ CONCEPTS / New- \ / cat = input \
| _______ | | | | _______ ( Concept ) | |
| /old \!!!!|!!|!| | / Old- \ \_______/ | C match! |
| / image \---|----+ | ( Concept ) | | +A match! |
| \ fetch / | |C| | \_______/----|--------|----T recog! |
| \_______/ | |A| | | \ | | |
| | |T| | __V___ \ | | C match! |
| visual | c|S| | / \ \ | | +A match! |
| | a| | | (Activate) \_V____ | ++T busy |
| memory | t| | | \______/ / \ | S skip |
| | c| | | | ( Parser ) | U skip |
| reactivation | h| |m| __V____ \______/ | P skip |
| | | |i| / \ |noun? | |
| channel | | |c| (spreadAct) |verb? | C match! |
| _______ | | |e| \_______/ |conj.? | +A match! |
| /old \ | |_|_| / ______V____ | R stop |
| / image \ | / \/ / \ | |
| \ store /---|-\ Psi /------( Instantiate ) | |
| \_______/ | \___/ \___________/ | |
See the
Activate
algorithm in
AI4U Chapter 33.
A copy of
AI4U, or a simple depiction of its cover, could be on
display as part of the interactive hands-on AI Mind exhibit.
A note on
AI4U secured in a plexiglass container could advise
museum-goers to visit the museum gift shop to examine a copy of
AI4U, or to ask any available docent to let them examine
AI4U.
Once museum patrons enter the gift shop to look at
AI4U, they
may find a selection of other books on artificial intelligence.
First a computer-savvy staff member will download the underlying
Win32Forth programming language and the
MindForth free AI source code.
Decision-makers at the museum will ask for a demonstration of the
software to determine if the AI Mind is advanced enough and interesting
enough to go on display at the museum. MindForth did not become
museum-worthy until
3.SEP.2008, when a new feature of
KB-Traversal
(knowledge-base examination) made MindForth vastly more interesting
and engaging to human users by reactivating latent concepts during
lapses in the thought-stream of conversation.
KB-Traversal solved
a chicken-or-egg problem. If the AI was interesting only when it
was visibly thinking, and if it was visibly thinking only when a
user was interacting with it, how would anybody start interacting
with the AI Mind in the first place? And if a few users now and then
did engage the Mind in conversation so that passers-by stopped to
watch what was going on, what incentive would there be for another
visitor to engage with an AI program that was just blankly sitting
there and not visibly doing anything?
KB-Traversal spiced things up.
The new
MindForth never stops thinking and throwing out ideas for
a museum visitor to respond to.
A science museum should not cheat the public by displaying a mere
chatbot full of canned responses to pretend that an intelligent
conversation is occurring between the visitor and the computer.
Even with disclaimers that the chatbot program is not an AI,
the intelligent museum-goer will wish for a more exciting exhibit,
something truly challenging to the human mind -- an artificial Mind.
MindForth delivers the real McCoy, True AI, and dares the user to
prove otherwise. The scuttlebut will drift around town that the
local science museum has an actual installation of the program
claiming to be a true artificial intelligence. The villagers with
their pitchforks will march angrily on the museum...no, just kidding.
Thoughtful types -- professors of philosophy, precocious students
from schools for the gifted, newshound reporters from weekly and daily
newspapers -- will beat a path to the door of the better AI mousetrap.
And if the
MindForth AI is still too primitive to warrant installation
as an exhibit, give it another year or two of improvement and
IQ-upping.
The very process of positioning the AI for adoption by museums may lead
to not just a few programmers building the AI, but to a vast army of
AI programmers taking up the challenge. To paraphrase Werner Heisenberg
in Das Unbestimmtheitsprinzip (The Uncertainty Principle), to observe
a phenomenon is to change the phenomenon. The more people look into
the True AI claims of
MindForth, the more people will either improve
MindForth or create similar artificial minds that may for some reason
be more suitable for installation in a science museum. As AI Minds
proliferate, AI display workstations may also proliferate in museums.
When the Feds raid the museums and close down all the AI workstations,
it will be too late. AI will be here, and the public will know about it.
Welcome to the Technological Singularity.
http://en.wikipedia.org/wiki/Technological_singularity
The AI Mind itself will be a very small part of the exhibit.
The bulk of the display will be the creative presentation of
background information on artificial intelligence and the
providing not only of answers to expected questions but
also of things that the public should consider thinking
about.
Since most of the display will be the posters and other props
surrounding the actual
AI Mind running on one or more computers,
exhibit designers and museum directors may wish to create images
of the (let's hope) award-winning, public-luring AI Mind exhibit
and videos of the interactions of people with the MindForth AI.
Check
http://www.flickr.com/photos/tags/mindforth/ to see if
photographers have uploaded mindforth-tagged photos of exhibits.
If you upload your exhibit images tagged with
mindforth, other
creative types will be able to find and admire what you have done.
The
kbSearch module was implemented on
17.SEP.2008.
It is currently the most sophisticated of dozens of mind-modules
in
MindForth. Although
kbSearch was originally created to enable
the AI Mind to answer questions tersely with "yes" when warranted
by the factual knowledge in the
knowledge-base (KB),
kbSearch is
expected to play a major role as the mentifex-class AI Minds
move into machine reasoning -- which requires logical testing
(and therefore searching) of conditional propositions in the
knowledge base. The
kbSearch module works by a kind of
NLP
thought-generation in reverse. It does not merely search for
words in the
knowledge-base. Instead, it conducts a dynamic
search for chains of thought based on
spreading activation.
The
kbSearch module actually activates the concepts which it
is asked to search for, in the order in which it is fed the
concepts. As long as the chain of activation matches up with the
search-query (e.g., "Do robots help people?"), the AI Mind
literally knows that the necessary conditions for a "yes"
answer are being met. If the chain of
spreading activation
wanders away from the idea being expressed in the search-query,
then the Mind rightfully fails to arrive at a "yes" answer, and
makes some other response or perhaps no response at all.
This powerful technique of not statically and not passively but
dynamically searching a
knowledge base is the technically
most advanced feature of
MindForth artificial intelligence
and could be the engine driving a dazzling
AI Mind exhibit in
a science museum or any showcase of artificial intelligence.