ainow

Open full view…

What makes this project special?

artficialmind
Sat, 07 Jun 2014 18:50:40 GMT

Hello there, I am very curious about your project. However I can observe that the general perception of your project is rather sceptical (because of some extraordinary claims you make). You have already provided some material in the Google docs area, however I think it would be a great idea to provide some major information in this forum (or better on the site) in a shorter and more clear form in addition. Can you please make bullet points containing answers to the following questions? 1) What makes your project special compared to other ideas and already existing projects? I mean there are already some projects like OpenCog. Why do you think your project will be more successful in such a confident manner? 2) In addition to 1) what big problems in artificial general intelligence research do you think you have solved that were unsolved before? 3) Why do you think your project will be implementable without any impediments? 4) Are there some claims/ideas that you can already proof in any way? 5) What is the exact expected outcome for the year 2016? If you want to promote this project it is not enough to provide some papers in Google docs. You should rather try to sell it by convincing people that you have a fresh, interesting and solid approach to such a big problem. The above questions and answers could also be integrated into a "frequently asked questions" section. I look forward to your response. Thank you and best regards artificialmind

OliMpele
Wed, 11 Jun 2014 09:10:18 GMT

first off: I like your idea about an FAQ. will definitely work on that ASAP. 1) it's the first project (AFAIK) that claims to have derived a complete and implementable software design. opencog etc. are still in the research phase - but I am in the construction phase. --> this project is further along than any other (public?) strong AI project 2) the NLP problem for one thing. but probably various others. I'm not familiar with the list of "unsolved" problems. also, same as I said here: http://www.reddit.com/r/artificial/comments/25z4pn/i_believe_i_have_solved_ai/chm4zi3 --> do you have a list handy of such problems that I could look at? 3) i've already written a good amount of exploratory code (over a years worth) in this paradigm, in different languages, and i've not encountered any "problems" - except the usual. so my intuition is, I will not encounter any problems. I have the same confidence in all the major parts of the design, as in the parts I have already written. I didn't write them in some specific order from most to least complex - only following the order of dependence between the components, and just arbitrarily focussing on one thing or another at a time, while I was still researching the rest. I don't see why one part of the design should have problems that don't pop up in the other parts: all the sub-algorithms use the same graph representation, and apply the same kind of transforms. It would really surprise me, if there was a hidden mechanism that would make it so. There is no complexity explosion in the runtime behavior, as best as I can tell. --> in other words: it works on paper. 4) I can illustrate solutions to most things, but it's best to do it in the code directly. what one needs to get a sense of, I think, is that solutions are possible in this paradigm. I will try to present some examples from various domains of thought, and how they are solved by the system at runtime.. --> hopefully I can come up with some diagrams that will make it intuitive. 5) if I get the funding I need, we'll be achieving the Singularity / Transcendence around that time.

ForauntO' connely
Tue, 24 Jun 2014 14:17:27 GMT

> if I get the funding I need, we'll be achieving the Singularity / Transcendence around that time. You propose to release a fully functioning intelligence without first testing it within a limited environment? This is incredibly risky, even if you are capable alone, of doing what has eluded an entire field of scientists, programmers and researchers for well over 40 years. I think many people would be encouraged to fund you if you could show a proof of concept AI system displaying intelligence inside of a constrained environment.

OliMpele
Tue, 24 Jun 2014 14:19:16 GMT

@ForauntO No, we intend to finish the development in a silo, with a moving wall on the code. However, this will not be of an issue until much later. We want there to be international competition about reaching the finish line.

ForauntO' connely
Tue, 24 Jun 2014 14:29:44 GMT

A race to completing AI sounds even riskier! if there were large stakes for the participants (and they are the largest stakes!) this might encourage the taking of shortcuts.