If we remain on target with 2012 as the expected arrival of the Singularity, we have only two or three years left for spreading the
SeedAi because it spreads as easily as germinal spores and then germinates as a living, thinking AI Mind on every host computer dedicated to open-source artificial intelligence or in every science museum exhibiting True AI to the public.
For the JSAI to be truly open-source AI, we need not only to publish the Strong AI source code but also to disseminate the developmental record and history of how the AI Mind came to be -- the ontogenesis as it were, or as it shall be in the unfolding of the moving-finger Cyborg archives.
If an AI user anywhere in cyberspace chooses to update the JSAI by following a Control Panel link to a newer version, that newer version will have an AI Lab Notes link to its own archival record of its own ontogenesis. In that way, the AI Mind and its archival labnotes will march forward in a kind of lockstep through time, with any JSAI release on a computer anywhere always linking to its own AI "birth certificate." Each AI update automatically points to a new "birth certificate." A nation, corporation or individual desiring to study the AI and perhaps reverse-engineer its functionality, will be able to track the emergence or disappearance of any AI feature over time by maintaining a repository of successive releases and their corresponding AI Lab Notes.
Whosoever creates a branching divergence from the original AI stem line is encouraged to at least keep (and consider publishing) some form of AI Lab Notes to record the developmental history of the divergent branch in AI evolution. In your own AI coding or AI Manhattan Project, the designated AI archivist is welcome to copy the methodology instantiated here in the Cyborg AI weblog. For instance, each AI journal entry shall typically correspond to a release of the newest AI source code. If the AI coding is prolonged over several days before a new release is issued, each day of work should be recorded not as its own blogpost, but as a subsection of the culminating blogpost made in conjunction with a release of the newest AI software version. In that way, there is only one cumulative journal entry for each release of the AI software, even if the work of the AI coding spans a series of work days.
After each release of new AI Mind source code, the programmer or designated archivist initiates a new accumulation of AI Lab Notes, because the moving finger writes, and having writ, moves on.
Monday, August 31, 2009
Monday, August 24, 2009
This reminder is prompted by the recent appearance of
singularitarian stirrings in the ivy-coated halls of Academe. At the otherwise party-school University of Pennsylvania, for instance, plans are underway to offer a course on the Singularity with a published Tentative Class Schedule. Unfortunately, the Working Bibliography shows an extreme skepticism towards the Mentifex AI project, as if it were a reason to be wary of Singularity enthusiasts. Just wait until AI Superintelligence takes its rightful place on the most powerful supercomputer in every advanced economy. The moving finger writes and, having writ, moves on.
Friday, August 21, 2009
Wednesday, August 19, 2009
Cyborg weblog and I noticed that "Odna Mona"
was the very first Follower of the weblog.
Thank you for Following a fellow cyborg.
It was strange that I clicked on "Odna Mona"
but it took me to no information about her.
Maybe "Odna Mona" is a new account that has
no profile associated with it.
Yesterday I was intending to set up Google AdSense
on the Cyborg weblog to try to make a trickle of
money from it. I was technologically not able
to complete the process, because the Ruby Internet
cafe and the Ruby library did not accept "cookies."
The AiLab computer with Windows 95 is too obsolete
to even log in to the Cyborg account. So I may
have to try again from the fancy computer of
"Matthews" and his wife "Ballerina."
But maybe I should not even have Google AdSense
on the blog. I have been thinking that maybe I
should just go ahead and publish the blog and
not worry about making money from it.
I am about to start a major, complex campaign in
the AI programming. It involves "self-referential
thought" on the part of the artificial Mind,
that is, the ability of the AI Mind to think
about itself and ask questions about itself.
As it is, the Mentifex AI is barely able to
answer the question, "Who are you?" When it
answers the question, it is primitively engaging
in self-referential thought. It thinks,
"I am (whatever)." It was extremely difficult,
and took decades, not years, for me to bring
the AI to the point of such primitive thought.
But now recently I have been realizing that
the AI Mind needs additional concepts and
English words (what; where; when; why) in
order to broaden and fully develop its
intellectual powers of self-contemplation.
We need to be able to ask the AI the question,
"Where are you?" and have it tell us not only
its location (e.g., "in the computer") but also
our own location (e.g., "out in the world").
We then need to make dialogs possible in which
the robot AI tries to learn every possible fact
I have been noticing that there are webpages and
wiki-pages devoted to self-referential AI, and so
I feel that I am about to conquer some important
territory in AI expansion. Unless I am totally
misguided, of course. If I achieve good results,
I want to go to those various wiki-pages and
write up the AI results there. Wish me luck!
Tuesday, August 18, 2009
artificial intelligence (AI) in general, it is
being repurposed specifically for publishing and
archiving the MindForth Programming Journal (MFPJ)
in connection with the Google Code MindForth project
in open-source AI for robots. We tried to obtain "AI"
as the URL designation for this weblog, but "AI" was
already taken. We tried again with "mind" but it, too,
was taken. We tried "cyborg" for cybernetic organism,
and we obtained it -- back in 2001, the year of the
Our blog about AI programming involves a wide range
of topics as enumerated in the list of links associated
with this blog -- from artificial intelligence on up to
a supercomputer on which we hope eventually to install
versions of our AI software. Our AI is meant as the
brains of a robot in control of sensor input and
actuator output. The design of our AI is based upon
linguistics and neuroscience.
We hope that Google AdSense advertisers will be
pleased to see their ads show up on our humble weblog.
Our audience will be small, select and high-strung,
because the dry and technical details of programming
artificial minds will probably not appeal to hoi polloi
surfing the Web for kicks and high-jinks. Probably only
decision-makers with large sums of discretionary tech
funds will bother to scrutinize our obscure AI blog.
Agents of industrial espionage may bookmark our blog in
order to scope out and scoop the futuristic AI landscape.
Nerdy geek types who are foolish enough to spend all their
money on building amateur robots -- they are among our
intended audience. We are writing for Joe Appcoder,
not Joe Sixpack. Joe Ivy in the philosophy department
might like us, too, but not Joe Couch Potato. We are
boring, dull, and of relevance only to the gleaners of
every tidbit about the coming Technological Singularity.
Friday, August 14, 2009
=== Disclaimer ===
This wiki-page discloses the existence of MindForth Programming Journal
(MFPJ) files and it links to on-line archives of MFPJ files, but it does
not itself contain actual MFPJ entries, for several reasons. Although an
MFPJ entry is part of the ultimate documentation of MindForth AI, it
only documents how MindForth came to be, and not what MindForth is at
its currently most developed state. Any attempt to record MFPJ entries
here at the Google Code MindForth (GC/MF) site would fail to be complete
and comprehensive. The earliest dozen or so volumes of MFPJ notations
exist only on paper and not in electronic or photographic form, and so
only later entries, and not all entries, exist as electronic files.
Some MFPJ entries were originally published on UseNet and other venues.
A goal for this wiki-page is gradually to link to as many MFPJ entries
as can be located and/or organized on-line.
=== Caveat ===
Since pages of the MindForth Programming Journal may or may not be
interesting to AI enthusiasts who wish only to evaluate the AI
Completeness of MindForth and who are not likely to be interested
in every little decision that was taken in the coding of the AI,
these MFPJ entries are organized and listed here for the sake of
providing too much information rather than too little. The MFPJ
files are not blared and trumpeted for people to look at. Rather,
the files rest quietly at diverse locations for perusal by true
AI enthusiasts of True AI. If someone wants to track down the
genesis of a feature in MindForth, these files may shed some light.
=== History ===
The MindForth Programming Journal (MFPJ) began in 1995 as a
handwritten record on paper of the efforts of MentiFex both to
learn the Forth programming language and to port MindRexx AI
into Forth on the Commodore Amiga 1000 computer. After an
initial flurry of activity, MindForth and its MFPJ diary
languished until the start of 1998, when the coding of
MindForth resumed truly in earnest and in response to
challenges encountered by MentiFex in the Inferno of UseNet.
Over the next decade, MFPJ entries were published sporadically
on-line as UseNet posts, usually to announce some advance in
the AI coding. A Google search on "MFPJ" may yield results.
=== Publication as Web Pages ===
For several months in 2008 on a trial basis, MFPJ entries
were published on-line at the following URL locations.
The files listed above as clickable links were created by
alternating between composing lines of AI source code and
documenting the process of coding in the MFPJ HTML file,
which was then uploaded to the Web. When the process of
coding became quite intense, sometimes only scant notes
went into the MFPJ file until satisfactory results were
achieved and recorded as <i>fait accompli</i> developments.
Along the way, it became obvious to the AI coder and
MFPJ diarist that each published page of the Journal was
indistinguishable from what Jorn Barger was the first to
call a weblog -- although he did not earn a dime from
inventing weblogs. Thanks to his efforts, though,
unto MentiFex may accrue some AiFunding if the MFPJ
actually becomes a weblog with advertisements interspersed.
Since each episode of MFPJ writing and MindForth AI coding
culminates in the release of a new version of the AI Mind,
there is plenty of opportunity to include weblog analogs
in the early stages of gathering ideas to commence coding.
All manner of BTW (by the way) links may be inserted into
the brainstorming portions of an MFPJ-as-weblog entry.