James Curran - Apr 29, 2014:

Et al,

*** FLASH *** SIMPLEX FIRST PUBLIC DEMO *** FLASH ***

... the conversation appearing on the screen should be read out by two people, one playing the part of the user and one playing the part of the computer.

Am I understanding from this post and a former post about an ability to ask itself questions, that the demo displayed a conversation of the bot with itself, consisting of {scripted | auto generated} questions each followed by the bot understanding the question and formulating a response, in-situ, using one or more implemented reasoning techniques e.g. {reduction | first order logic | rules | belief/Bayesian | competition/satisfaction | generalization/specification | temporal | spatial | circumscription | consistency | frame/slot pattern matching | planning ... } ?

(I apologize for jumping into this thread without so much as having introduced myself.  I promise that will follow, but my curiosity has been piqued to know more specifically if there is something here for me to learn by investing in reading and thinking about this thread. )

With great sincerity,

Alan

Sincerest apologies to all on this thread for discussing a subject which may be easily misunderstood. Strong feelings, yes, but for good intentions not bad (See: disclaimer promise two posts back).  Some members such as advanced auto mechanic, Dave, industry certified, such as we, and/or natural language processing,  information systems college grads, have actually made a living maintaining artificial intelligence in the field —Skip to the point, that good solid A.I. actually in service in the real world isn’t glamorous by today’s standards.  So it just doesn’t get talked about much, doesn’t nearly get the credit it deserves. 

A.I. is more about vintage works of genius, than modern day web based standards. Designed long ago on real time Y2K noncompliance, computer languages long gone, such as PL/M which predates C Language.  Practical A.I. has been and is still around controlling elevator banks and oil burners in city buildings since before even operating systems, such as DOS became widely available. 

Prototyped with wire bundles to individual pins of industrial CPUs, resembles a lock of golden human hair. Salaries so high they’re off the payroll charts, but companies have little choice. Dirty little secret rarely discussed is that most vintage A.I. works of genius can’t be reproduced by modern standards without a very significant investment. They wind up embedded “as is” on drop-in replacements as integrated circuit chips. 

Well, whenever anyone comes in from that era—It’s an A.I. discussion worth having.

∞Pla•Net - May 28, 2014:

Well, whenever anyone comes in from that era—It’s an A.I. discussion worth having.

I remember the days when you had to design your own keyboard and you could store your programs on punched paper tape. Heated discussions would revolve around whether or not to add a dedicated key for a function…

Hype vs Hope

Most of the members of this forum are AI/chatbots enthusiasts. We look forward to new discussions of methods and technologies that can spur ideas and conversation. We look forward to the unveiling of a new bot or algorithm with the same zeal that comes from the aroma of an excellent home cooked meal. The anticipation increases the hunger and you can’t wait to be satisfied with the meal. But, if the meal never shows up, all you are ever left with is an empty feeling. We Hope you have something new and exciting (and that it is not just Hype).

Many have valid reasons to keep their creations close to the vest. Hans Peter Willems has gotten funding and is building a company based on his theories. Mark Atkins is looking for funding (and looking for a tenure track professor) and doing a proposal. But, both found ways to contribute to the discussion without revealing too much.

But, the problem with GTSSB* claims is that they often have had little merit. Chatbots.org has had its share of people who claim to have the next big thing and unraveled the secret to strong AI (and even published dates and delivery times) only to disappoint. So, the board has grown skeptical as pretenders come and go. I now adopt a “The proof is in the chatting.” approach.

“James Curran”:
My unstated hope was that it would allow me to provide you and/or those attending with a demonstration of Simplex and provide me with the opportunity to seek advice on how I should proceed in presenting something so radical to the world. I also lack the technical knowledge to confidentally define what I have developed.

Hopefully, someday you will be able to demonstrate Simplex to members of the community and they will be able to report back and enable us to give you better feedback. If all else fails, I would suggest attending the Loebner Contest (even if you do not participate). There will be a number of members there. It would be a good place to gain interest.

*Greatest Thing Since Sliced Bread

author of the most advanced AI system / chatbot unless you tell me otherwise!

James, let me “tell you otherwise” by rephrasing descriptions of Skynet-AI and its technology into terms similar to what you are using.

James Curran - Apr 13, 2014:

If Simplex is the only system that can ask itself questions then in my view it is significant.

Skynet-AI can also ask itself questions, and it can have entire conversations with itself.
One of the design features of Skynet-AI is that it can be used as though you were chatting with a friend, or it can be put into a mode where it chats with itself!

When it fails to do what is expected or has been asked (in plain English) then I perceive that occurrence to be a bug and I intervene by perhaps defining words or phrases that are not known or are mis-understood. Sometimes there is need to analyse where and why a fault occured. Obviously, the end user is not expected to deal with such a failing! But this also never requires a C programmer.

If the fault requires a change, no one ever touches C language code.
JAIL(TM) (Javascript Artificial Intelligent Language) was a framework that I invented to explore AI and chatbots. I tried to think about what things computers can do really well and bridge natural language to them. Skynet-AI is a demonstration of the underlying JAIL technology.
The potential of JAIL is not in the static demonstration of Skynet-AI, but in what it could be used to create.

JAIL can be all things to all people who speak English.
>You don’t need more than English with the ability to do real-time translation.
http://blogs.technet.com/b/microsoft_blog/archive/2014/05/27/microsoft-demos-breakthrough-in-real-time-translated-conversations.aspx

Yes ...  JAIL can be used as a programming language. The user only needs to use English.

In developing JAIL a purpose designed language has been developed with the ultimate goal of facilitating communication and programming using plain English. No compiling is required.

Skynet-AI is an integrated demonstration system that does not use third party software such as “Mark Up Languages”, dictionaries, data bases etc. Accordingly, it does not suffer from cumulative third party limitations imposed by such components.

JAIL could be a game changer in the field of information technology because it doesn’t seek to maximise the “Eliza Effect” but is used in developing the building blocks of language comprehension. Skynet-AI does try to use the “Eliza Effect” to its advantage.

Skynet-AI has hundreds of functions but uses its capability to understand English to overcome the problem of complexity. In other words, the user tells Skynet-AI what he wants to do and Skynet-AI responds by running the function needed.

In the future all software could concurrently run JAIL in the background and users will then use NLU (Natural Language Understanding) as and when needed to achieve what they are trying to do.

When the author is wearing stronger rose tinted glasses he sees a world that uses JAIL as its preferred computer language to efficiently exploit conversational user interface capability and its many other benefits which can include:
- The basics can be learnt in minutes
- Easy to use
- Easy to maintain
- Shorter development time
- Can be used by laymen as well as the professional programmer
- Never needs compiling

All user interaction with Skynet-AI takes place through plain language. In other words Skynet-AI uses its ability to understand English to do what is required. This means that the program is continually testing itself! In the event of a failure of understanding, it is a simple process to teach the AI how to correct itself.

Skynet-AI currently runs on hundreds of platforms. It is know for flexibility and speed. Feel free to have a chat.

Alan McDonley - May 28, 2014:

(I apologize for jumping into this thread without so much as having introduced myself.  I promise that will follow, but my curiosity has been piqued to know more specifically if there is something here for me to learn by investing in reading and thinking about this thread. )

You’re welcome here, Alan . Those are intelligent questions. Having read and thought more about this thread than I should ever have, and being interested in all the facets you mention, I can advise not to read this topic. I’m usually one to come up with great ideas from the silliest posts, but I haven’t had a glimpse of an idea spark from this entire thread. James has been adamantly vague, isn’t willing to divulge how his system works, and prefers not to bother with any distractions like learning about the methods you named. I’m afraid this thread is bare of answers, but I’m glad to say it isn’t representative of this forum.

I’ve exhausted every advice one might give within decency. I’m just glad that James isn’t a spokesperson for my system, which by the way is at least a match for his, but I’m not willing to back that claim up . James isn’t obligated to prove anything to any random person on the internet, but if one can’t make a plausible case to a less informed public, yet expect AI professors to go out of their way for it, suit yourself, really. I have NLP to program.

8pla: I’ll get back to you on your interface there. I’m interested.

Don Patrick - May 28, 2014:

I can advise not to read this topic. I’m usually one to come up with great ideas from the silliest posts, but I haven’t had a glimpse of an idea spark from this entire thread.

Worth reitterating- this thread is an intellectual cul-de-sac.

Hi Alan,

Welcome to this site.

As a fellow newbee I have found it to be a steep learning curve.

One of the astonishing facts that I’ve had to learn is that people do not take you
at your word.

This is something of a problem to somebody like me who simply wanted to let people
know that there is a new kid on the block. I’m of the age when “My word is my bond!”
Seemingly, this is no longer the case.

It is not sufficient to say something like “The screens I have described are a true
and fair representation of what happens.” People are apparantly upset if you do not
provide an explicit account of the underlying logic. As I am commercially naive, I
instinctively feel that it might be foolish to place something into the public domain
that is the result of years of private research. I simply do not know what is
unique and so commercially sensitive and what is common place. I have pointed people
in the general direction of what I believe to be the way the way to go but that
appears to be not enough.

It is not compulsory to read this thread so I find it difficult to understand why it
should be so disturbing and why those who consider it to be an intellectual
cul-de-sac ... should still read it!

Everything I have said I have believed to be true at the time of writing. Having said
that I now feel guilty about not prefixing my post with the title “WITHOUT PREJUDICE”
and adding the caveat at the end E&OIE; ie errors and ommissions excepted. I’m left
feeling shell shocked by the tenor of some responses that imply that all I’ve said
will be used in evidence against me!

That does not encourage further revelations about Simplex’s capabilities.

One of the most suprising consequences of joining this forum is the amount of time I
have spent in thinking about and dealing with questions etc. Also the feeling of guilt
at not saying thank you to the occasional member who seem to understand my approach.
This is making me review how my time should be spent.

As regards your question, only you can decide whether it is worth your time! The number
of my posts are relatively few and so are not too onerous to read.

I apologise for my style of writing which also seems to be a cause for concern to some
members.

I should add; the fact that I know so little about other peoples work does mean that
I approach AI problems with fresh eyes. I suspect that occasionally this is beneficial,
even if it does mean that at times I rediscover a wheel! It also means that I am not
tempted to criticise other peoples hard work.

Nevertheless, I feel I have a grasp of current AI abilities and that alone encourages
me to continue with with my current approach.

Jim.

Don Patrick - May 28, 2014:

8pla: I’ll get back to you on your interface there. I’m interested.

Oh, that’s great to hear… Thanks for your interest!  I look forward to it.

James Curran - May 29, 2014:

I find it difficult to understand why it should be so disturbing

In very basic terms, topics like this translate to “Mine is way better than any of yours but I’m not gonna show it”. It undervalues other people’s hard work, even if you didn’t mean to. People who believe in their own work as strongly as you believe in yours.

Actually, coming from a different background, I see the same thing happen frequently on comics fora: Someone will post “I have written a hit story and I’m looking for a publisher.”, but they are never willing to post even a paragraph of the script for fear of copycats. And so no-one is able to judge how it compares to other stories, especially potentially interested publishers. Merlin’s explanation is well worth the read.

Mr Curran,

Maybe I can help put some perspective on some of the views expressed that might help. Nowhere in the entire realm of academia will you be taken at your word, at least not that I am aware of. If you wish to be taken seriously, you must publish. I agree that quite often, creative thought blossoms outside of the usual channels, however it is still required to follow the same rules as those that bind individuals that are working in large corporate or academic institutions. That is you must define a hypothesis,  construct a proof, and prove or disprove your work. Then relating it to the work of others in the field, publish it in a form that can be followed by those in the field. It does not have to be made public in a general sense, you can write white papers privately and have them reviewed under the shield of a non disclosure, etc.. HOWEVER, you will still have to publish something that displays at least some knowledge of the field publicly. As this relates to the reaction here, consider this. The attention of the world is focused on the field of AI right now, academically but in particular the tech sector of the business world, and you can divide those involved into (2) basic groups.  Large companies and academic entities, and smaller businesses and individuals. As small businesses and individuals we are competing for funding with entities such as Apple, Microsoft, MIT, etc…As I mentioned previously, we are expected to follow the same rules as everyone else under great scrutiny, so if as a group we were to throw that to the wind and accept something at face value, it lessens our own credibility in the field. Much like a serious cosmologist who suddenly announces that he believes in astrology because a psychic told him that his stars were aligned.  Now I am not saying that your work is not real or valid. its just that the language that you are using does not reflect that you possess an underlying knowledge in the field.  I am not saying that independent breakthroughs do not happen outside the regular academic channels, in fact they are probably MORE likely to happen there. Albert Einstein is said to have remarked “it is a miracle that creative thought survives formal education”. However now that you have a hypothesis, it is incumbent upon you to research how your work relates to the work of others so that you can converse fluently in the field. Academic credentials do not guarantee that someone is producing independent or even relevant work, but they do demonstrate an underlying knowledge of the subject matter, so without credentials you will have to work harder.  My own credentials are rather lightweight, but I have a verifiable body of work in the form of work history, robotic systems that I have designed and built for various companies, as well as several patents and published materials in various fields.  My suggestion to you if you wish to be taken seriously at the same time protecting your core ideas would be to write some satellite papers which demonstrate that you understand the material, in a form which is acceptable academically, that is to say create an abstract, a body, a summary and annotate it correctly, and publish them. To be specific as it relates to your postings, you use the term SIMPLEX which is a term known in the art. http://link.springer.com/chapter/10.1007/978-3-642-29347-4_13#page-1
Which makes me wonder whether or not you are working with some form of ANN, you mention using NLP to create programmatic interpretation of data, this is known in the art, RICH uses something similar, MIT is working on something along the same lines, but you are not using language I can relate to. You mention SIMPLEX asking itself questions, and most of the guys took this to mean that it was talking to itself, I thought you meant some type of subsystem which is attempting to relate (or ground) concepts in the background.  Again we are not relating to the language you are using. Now if you had said that your system is learning through some stochastic process involving Bayesian algorithms, we would have related to things we already know.

Anyway, back to the reason for the post, nowhere in the entire world of academics that I am aware of, is your work expected to be taken at face value, and to do so would lessen the credibility of the accepting individual in the eyes of serious academics and certainly potential investors.

Sincerely,

Vincent L Gilbert
CTO RISOFTDEV inc.

QOTD

“It does not matter how big a fish you are, you swim in the same ocean with the same sharks as all the other fish” V. Gilbert

Excellent piece Mr. Gilbert! 

I would suggest your piece be published as a front page article on Chatbots.org, or perhaps submitted as an article to the Communications of the ACM magazine, where ELIZA was published originally in 1966. 

Joseph Weizenbaum published the article, “ELIZA — A Computer Program For the Study of Natural Language Communication Between Man And Machine” in an article for the CACM, the Communications of the ACM (Association for Computing Machinery [founded in 1947]).

CACM magazine is for readers with credentials in computer science and information systems.  What does this classic ELIZA article by Joseph Weizenbaum mean for computer science (CS) or information systems (IS) college graduates in terms of credentials?  It may sound a bit amusing to wonder whether these college degrees may be considered credentials for chatbots?

Don,

Please remember, Eleanor Roosevelts’s observation: “No one can make you feel inferior
without your consent”.

Jim.

Vincent,

An excellent post.

However please remember, George Harrison ... a humble carpenter and clockmaker who
solved the longitude problem that bafffled the most emminent scientist of his era.

There is still room for people who say ... “It does what it says on the tin!”

Jim.

- Wow, looks like I missed out on a good thread here. It’s a bit long and I don’t have that much time, so I haven’t gone though all of it, my appologies…
- @Vincent: before I read your post, I was thinking of saying something similar. And you are right, if you can, publish about your work. If you are more like me, and not that good of an accedmic writer (or no links to universities or something), than it’s even more important that you show what you already have. How else can people evaluate what you have done?
- @James:
Perhaps you can see your work as a stepping stone: show it to people while you are talking about it, let them see what you can do. If you really are good, people will be impressed and eventually, the right people will come along (I read in some of your posts that you consider yousrelf ‘commercially naive’, so you’ll need others to handle that part)...
Don’t be affraid about people steeling your ideas. Your value lies in the fact that you can create such things, not those 1,2 or 3 things you have already created. If you really are good, ideas drop out of your sleef like water from a waterfall and implementing them is only a matter of picking the right one to work on…
Of course, that inties that you are able to drop what you already have in order to pursue something that could have more commercial values.
If you don’t believe something like that can happen, that’s how I managed to become lead R&D…