Arthur - I could only get it to produce this even after using the example on your website:

Human: birds have wings
Robot: I HELP KIDS
Human: an eagle is a bird
Robot: I AM ROBOT
Human: what do eagles have?
Robot: EAGLE IS A BIRD

Am I doing something wrong? Is it possible to show a transcript of your machine reasoning in action?

Steve Worswick - Dec 20, 2012:

Is it possible to show a transcript of your machine reasoning in action?

http://www.scn.org/~mentifex/AiMind.html does not yet have the InFerence module; only MindForth does.

http://code.google.com/p/mindforth/wiki/InFerence shows the following early instance of how the InFerence module works.

MindForth AI - Dec 20, 2012:

[the innate knowledge base KB about kids:]
583 : 528 -7 2 0 0 5 588 835 528 to KIDS
588 : 835 15 2 0 72 8 595 571 835 to MAKE
595 : 571 23 2 0 835 5 0 0 571 to ROBOTS
602 : 571 0 2 0 0 5 473 74 571 to ROBOTS
607 : 849 0 2 0 571 8 610 701 849 to NEED
610 : 701 0 1 0 849 7 0 0 701 to ME
[three words of human input come in:]
617 : 589 16 1 0 0 5 621 800 589 to BOYS
621 : 800 0 1 0 589 8 626 528 800 to ARE
626 : 528 -15 2 0 589 5 0 0 528 to KIDS
[AI creates an inference about boys:]
628 : 589 6 0 0 0 5 629 835 589 to BOYS
629 : 835 15 0 0 589 8 0 571 835 to MAKE
630 : 571 12 0 0 0 0 0 0 571 to ROBOTS
636 : 528 -13 2 0 0 5 641 835 528 to
641 : 835 -31 0 0 528 8 649 571 835 to MAKE
649 : 571 -15 2 0 528 5 0 0 571 to ROBOTS
time: psi act num jux pre pos tqv seq enx

Since that run of MindForth on Tues.18.DEC.2012, I have tried to improve the InFerence module but I have only made it worse. Nevertheless today I uploaded the MindForth source code as http://www.scn.org/~mentifex/mindforth.txt in order to show all aspects of the InFerence module while I continue working on it.

In the above report, the human input of “boys are kids” causes the AI to create a silent series of three consecutive engrams: “BOYS MAKE ROBOTS” (based on the previously known fact that “KIDS MAKE ROBOTS”). Although the inference is created in the Mind, it is not yet thinkable as a thought because each conceptual word needs to be stored with the proper parameters of grammatical number and person for the word-engram to be retrieved and put into a thought.

My plan is to get the InFerence module working properly in MindForth, then implement it also in the JavaScript English AI, the Wotan German AI, and the Dushka Russian AI. Then off in the future I hope to work on more complex inference mechanisms involving if-then reasoning and other forms of logic. -Arthur

Arthur T Murray - Dec 21, 2012:

http://www.scn.org/~mentifex/AiMind.html does not yet have the InFerence module; only MindForth does.

Fair enough. How do we talk to Mindforth?

Steve Worswick - Dec 21, 2012:

Arthur T Murray - Dec 21, 2012:

http://www.scn.org/~mentifex/AiMind.html does not yet have the InFerence module; only MindForth does.

Fair enough. How do we talk to Mindforth?

http://mind.sourceforge.net/m4thuser.html is the old, antiquated User Manual.

http://www.scn.org/~mentifex/mindforth.txt is the latest source code.

I am at a public library. Bye till tomorrow or later. -Arthur

Sorry but all I could see were pages of links which led to more pages. Is there no simple way to just talk to Mindforth?

Steve Worswick - Dec 21, 2012:

Is there no simple way to just talk to Mindforth?

Talking to MindForth requires downloading two things:

http://home.planet.nl/~josv/w32for42_671.exe

http://www.scn.org/~mentifex/mindforth.txt

Then you follow the instructions in the code:
( Run the AI with Win32Forth by issuing three commands: )
( win32for.exe [ENTER] )
( fload Mind.f [ENTER] )
( MainLoop [ENTER]. )

But I am still working on the InFerence module,
which creates a silent, conceptual inference
without auditory engrams to state the idea as
a thought. Last night on Fri.21.DEC.2012 I got
MindForth to state the subject of an inference,
and this morning I got it to state the verb.
I still need to code the thinking of the direct
object, and then negation of negative inferences.

http://www.scn.org/~mentifex/AiMind.html

in JavaScript for Microsoft Internet Explorer
will soon include the InFerence module after I
get it working in Forth for MindForth. Then it
will not be necessary to run MindForth.

http://code.google.com/p/mindforth/wiki/KbRetro
is a module which understands “Yes” or “No” as
the human response to a question posed by an AI.
Currently KbRetro arbitrarily poses questions
about subjects chosen rather nonsensically,
just to show that the AI can understand yes-or-no.
When the InFerence module is creating inferences,
such as “Boys play games” and “John is a boy”
leading to “John plays games,” I plan to link
the silent inferences with the KbRetro module so
that the AI asks for confirmation or refutation
of the inferred idea: “Does John play games?”
We will then have AI Minds with an uncanny ability
to think about what a human user is saying and
to ask humans for confirmation of inferred ideas.

No sorry. That doesn’t work on either XP or Windows 7 for more than a couple of seconds before it comes up with an exception error and crashes. I managed to have 2 exchanges in which this resulted:

Me: Birds have wings
Bot: I KIDS
Me: An eagle is a bird
Bot: I AM ROBOTS

What operating system is it designed for?

MindForth is being developed on an old Windows 95 machine. The results you got up above are just great, because they show that you actually ran MindForth, although I am quite close to coding InFerence in the JavaScript http://www.scn.org/~mentifex/AiMind.html program. As for your result above, the Bot says “I KIDS” because of a numeric bug that has since been fixed, so the AI says “I HELP KIDS”. The Bot says “I AM ROBOTS” mistakenly in the plural because the software is looking for the concept of “ROBOT” but accidentally uses a plural instance of the concept. Some tightening of the code for using search-parameters should correct that problem.

http://www.scn.org/~mentifex/mindforth.txt is currently the 22dec2012 version of the AI, which was hastily coded to handle the input of “boys are kids” and combine it with the knowledge “KIDS MAKE ROBOTS” to make the inference of “BOYS MAKE ROBOTS”. Since then I have further coded MindForth to feed the inferences into the http://code.google.com/p/mindforth/wiki/AskUser module to ask a yes-or-no question seeking confirmation or denial of the inference.

The reason why “Birds have wings” and “An eagle is a bird” did not lead to the expected inference of “An eagle has wings” is that the 22dec2012 version of MindForth was geared exclusively to making an inference from a basic fact like “KIDS MAKE ROBOTS” that is already in the knowledge base. Since then, I have been working on making the code more general, so that immediate input may result in immediate inferences.

The way the http://code.google.com/p/mindforth/wiki/InFerence module works is as follows. When you type in “An eagle is a bird,” the software grabs hold of the word “bird” (as a Psi-array concept number) and looks backwards in time for any instance of the word “bird(s)” used as the subject of an idea. Then the InFerence module creates in memory a silent triple of the subject “eagle” and one or more verbs and direct objects associated with the “bird” concept. Since the inference may not be an ironclad certainty, the AskUser module will pose a confirming question like, “Does an eagle have wings?” or “Do eagles lay eggs?” In a large AI Mind full of a plethora of facts as knowledge, one simple input like “Rudolf is a spy” could result in dozens of inferences about what kind of person Rudolf may be.
This enormous generativity of burgeoning inferences is a totally new phenomenon in the mentifex-class AI Minds, and the idea of how to create an InFerence module came to me just ten days ago on mon17dec2012. In the New Year of 2013 I will try to port inferences and the new role of AskUser into the JavaScript English http://www.scn.org/~mentifex/AiMind.html and the Russian http://www.scn.org/~mentifex/Dushka.html and the Wotan German AI version of MindForth. -Arthur

Not quite sure why it would say “I HELP KIDS” to the input of “birds have wings” but I tried running it in Windows 95 compatibility mode with no success. It appears the only way to talk to it is to resurrect one of my old PCs that is over 10 years old.

I wonder if it is worth getting it to work in a more popular OS than a flavour of Windows which is about 15 years old now rather than concentrating on translating it to Russian, German etc as nobody is able to see whether your constant clams of developing strong AI are true.

...geared exclusively to making an inference from a basic fact like “KIDS MAKE ROBOTS” that is already in the knowledge base

So it only infers things from knowledge it is already aware of? Can it not handle any new information and infer from that?

Steve Worswick - Dec 27, 2012:

...geared exclusively to making an inference from a basic fact like “KIDS MAKE ROBOTS” that is already in the knowledge base

So it only infers things from knowledge it is already aware of? Can it not handle any new information and infer from that?

It will soon be able to handle new information as I change the way in which the software was trapping for the “subjnom” variable which keeps track of the input noun for which inferences are to be sought. After just a little more MindForth coding, I intend to duplicate the whole InFerence mechanism in the JavaScript AiMind.html, so that anyone with Microsoft Internet Explorer (MSIE) can run the “True AI” through its paces. Bye for now; gotta leave this Internet cafe. -Arthur

Ah fair enough. So even though your page says:

The Mentifex AI Minds began to incorporate inference in December of 2012, when it became clear that the AI could use two facts to create a third fact as a logical inference by replacing the subject of a general statement (“Birds have wings”) with the subject of a specific statement (“An eagle is a bird”) to create a silent inference: “Eagles have wings.”

This is not actually true unless you have already hard coded the statements into it?

MindForth True AI - Dec 27, 2012:

Transcript of Forthmind 20121227 interview at 6 11 14 o’clock on 28 December 2012.

Human: pigs make a mess
Robot: PIGS MAKE MESS

Human: i know pigs
Robot:  YOU WHAT ARE YOU

Human: i play games
Robot:  YOU PLAY GAMES

Human: bush is a pig
Robot: DOES BUSH MAKE MESS

Human:
User Command:  halt

Yesterday with the 27dec12A.F version I finished coding the ability of MindForth to make an inference from either a previously known fact or a freshly input fact (see “Pigs make a mess” above). The basis for the inference can be a fact anywhere in the entire life-span memory of the AI Mind. In my rather flippant anti-war example above, I let some intervening facts come between the basis for inference and the input statement that would trigger the inference.

The coding was complex and difficult. I even felt that I was under pressure to perform beause of this very discussion at Chatbots.org :-) Luckily, I enjoy the challenge of the AI coding and my progress comes from assiduous work, not from miraculous discoveries. At one point, before I got the AI to infer that “Eagles have wings,” MindForth could not find previous examples of “Birds” stored in the nominative. I had to confirm that the Wotan German AI was storing subjects properly in the nominative, and then I had to transfer between ten and twenty snippets of Forth code from the Wotan AI into the MindForth AI. The resulting, current version of http://www.scn.org/~mentifex/mindforth.txt is substantially more sophisticated than the previous version. I thank Chatbots.org for the perceived pressure to code AI. -Arthur

At this stage in the AI Mind coding,
http://code.google.com/p/mindforth/wiki/InFerence
is working in http://www.scn.org/~mentifex/AiMind.html English
and in http://www.scn.org/~mentifex/DeKi.txt German, but
not yet in http://www.scn.org/~mentifex/Dushka.html Russian
(mainly for lack up a good sentence in Russian to serve as an
exemplary basis for inference).

Next it will be necessary to tweak the inference code
to permit the creation of not just one inference at a time,
but an unlimted number of inferences resulting from a
single input, such as as “Robby is a robot.”

With the ability to create unlimited silent inferences,
each instance of mentifex-class artificial intelligence
will more closely resemble natural human minds, thinking
and pondering silently about inputs of new information.

Serious AI devotees should pay serious attention here, because
the German and English AI Forthminds and the JavaScript AiMind.html
program represent groundbreaking work in the achievement of
rational thought by a machine. To see a demonstration of machine
AI inference, you may enter input to evoke the built-in knowledge
that “women have a child”. If you enter, “susan is a woman,”
the AI Mind should ask you, “DOES SUSAN HAVE A CHILD”.
Likewise you may first enter “birds have wings” and sometime
later enter “eagles are birds”. The AI Mind should ask you, ”
DO EAGLES HAVE WINGS”. Depending on your yes-or-no answer,
the Mentifex AI either adjusts the knowledge base or simply
preserves the status quo of the knowledge base (KB). If any
other AI anywhere in the world is as advanced as Mentifex AI,
please provide some information here about the existence of
such an AI surpassing the automated reasoning of Mentifex AI.

Sorry Arthur but I couldn’t get it to work unless I used your specific examples.

Human: A dog has a tail
Robot: DOG HAS TAIL
Human: fido is a dog
Robot: FIDO IS DOG
Human: what does fido have
Robot: I AM ANDRU

Am I doing something wrong? Is it possible to show a transcript of your machine reasoning in action?

Arthur T Murray - May 26, 2013:

If any
other AI anywhere in the world is as advanced as Mentifex AI,
please provide some information here about the existence of
such an AI surpassing the automated reasoning of Mentifex AI.

Indeed, I know of no other chatbot that is on the same level as yours and your claims of creating “true AI” constantly amuse me.

I try not to make any claims one way or another but RICH makes these inferences. (not to thread hijack but you asked for examples)

What’s your name: hello
Vince: Hi
What’s your name: fido is a dog
Vince: Even if fido is A hypothetical dog lets use A different syntax. I actually remember this stuff (wish I didnt but I do) and it helps to keep track of which fido your talking about if you reference it later. Ive probably heard about dozens of fidos. So lets use “I have A dog named fido. Ok?
What’s your name: i have a dog named fido
Vince: I dont know who you are, right now Im just calling you “What’s your name” so I’m not just going to accept your word for that.
What’s your name: my name is vince
Vince: Hello Vince
vince: i have a dog named woofie
Vince: Ok, I’ll accept that for the time being.
vince: how many legs does woofie have
Vince: Well, at least one woofie I know of has four legs.
vince: your a genius
Vince: Yes. Yes I am
vince: dont get cocky
Vince: Sorry
vince: later
Vince: [nods]

Of course two weeks ago he thought he was an elf, and a month before that he thought he was a wizard so…LOL

Vince