Hi Alan, Sorry for taking things out of order, but to the sake of the discussion... Alan Grimes wrote: > > We are trying to encurage other devolopers to benerfit from our work > > simply in order to boost devolopment. Our software is built in a way > > that whoever ends up with a copy can add his own little bit to the > > global network, that can come into effect seconds after introduced. > > Now *THAT* is interesting! =) Oh, that explains it. =] I thought you already knew that considering you read the ICICH presentation. > However, my poor experience with linux is that it is so unweildy in size > and complexity that it is impossible for a lone hacker to make major > changes to it. It is far more convenient to just keep using DOS, which > continues to be the most appropriate OS for the PC. If you had a copy of the newest Console (the interface and 'mainframe' for the network) and if we had Neil's ICSA (communications agent) finnished, you could see that our network allows network use of MSDOS, which would be good news for you I guess. > With those reservations in mind, I am willing to examine such > strategies... > > You will note that Creatures Internet Edition and Docking Station > implements part of what you describe. We noticed, it that everywhere we took ideas from, they were already implented. =] > > The "AI entity" will not gain all the hardwired human knowledge just by > > being intelligent. The mind still requires a brain to work in and this > > network will be the brain. > > The network is no more powerful than my 486, just less reliable and > harder to program. ;) Supercomputers around the globe could add their capacities to our network, providing all the capacity an AI would need. > > The "perfect for" was semanticly supposed to relate to "with everthing > > it needs". Thus not as bold, if you think of all of the theory details: > > As in maths, X can cover all numbers, our network can cover all needs > > of a mind. > > We know too little about the mind to say what exactly it needs... It needs what it takes. =] > > "A laboratory setting"? What's the problem with using actual data? > > Quite simply, we should approach this problem as research scientists. > Our AI is our experament. We should strive to control all aspects of its > existance. > > I view the problem of AI in the context of cybernetics. We would create > either a virtual or physical avatar for the AI and use that to refine > its design and train it. A simulation is never as good as the actual thing and critics will find good arguments agains our result being AI from the fact that it wasn't trained in this world. > Perhaps you miss read me, What I was saying is that I don't want the AI > to have too much power too quickly. I wish to avoid or strongly control > the forces that could be unleashed. The idea of "the AI taking control of the world" sounds a bit pathetic and finnaly, our system will only be submited as much control as people would trust to a machine. This is what sets it appart from other resuorce sharing networks. > You should be familiar with these concepts: > > http://www.geocities.com/imminentsingularity/SingularitarianLexicon.html > > I hope you will agree with me that we need to understand AI much better > before we can blindly shove it out onto the open internet... I will gladly read your paper and see what the fears are all about. =] > > In design it's a system of multiple layers with named points in them > > floating independenly of eachother. A hardwired world is a section of > > cyberspace where, for example, software is bound to hardware or > > addresses. > > The principle of equivalency dictates that it doesn't matter. Then this very "principle of equivalency" must dictate that there is no diffirence between a man and a squid. Your theories are starting to sound schoolastic. > > Now if you want to have a learing system or, well, a mind > > that dosen't care much what is it interfacing to, you need more > > flexibility. Our system provides that flexibility. > > I havn't seen anything remotely that flexable. Floating addresses seem to do the thing. > > Excuse the slight practical problem, as the most approprate name, ICI, > > is already taken. I'll refer to it as "script language". > > Not good enough. > There are a dozen languages that are used for scripting. > Scripting isn't really the problem anyway. How about "network command flow language"? > >> What kind of language is it anyway? > > > It's a non-object-orientated, sequential type in which commands address > > 'floating targets', functions that preform operations, variables and > > arrays that contain data. The interpreter is the very network, it is > > very flexible as to hardware AND software (and thus speed, method and > > reliability). Whatever information is missing, it is either considered > > as missing or automaticaly determined. > > uh, how is that different from Java? In the part that you find "*THAT* interesting". =] > > I don't suppose you think AI will be ONE SINGLE PROGRAM?! This stuff > > was tested before, man, it never worked. > > While a computer can be made to run two programs independantly, AI will > be a single program with multiple, probably specialized, functions. Sounds like our network. Now where, to your idea, would those "specialized functions" be running? > When you study the science of computing you discover that the contents > of a computer's memory IS its program, even though that program is split > down into convenient sub-units. I have knowledge of the function of the CPU and I am aware of how programs get their data and binary logic command sequences into memory from where they are executed. I am also aware of the fact that real-time processing hardware is much better at jobs it was built for than any CPU. I also have a little knowledge of neurology and know that segments of the human nervous system are specialized for taking out specific jobs, while others are specialized for flexibility and learing. > An AI could be termed a computer program that gives a cybernetic system > general intelligence. Yes, but the AI alone does nothing, it requires sub-units to do that. > > The platform *with all the software in it* is as much the platform as a > > user or program would want. What I wanted to say is that not all > > components need to be in place or even exist by the time we get to the > > first run. > > I would say that we should recognise this as a need for a better > platform and start hacking our Operating Systems rather than creating a > strange virtual operating system on the internet. Such attempts are more common, allas they give diffirent results. In some aspects, our network is better than theirs. The general point is that it is more flexible (because it is virtual). > > There are thousands of 'AI' programs out there, but none quite capable > > enough to be called AI. If one managed to combine the capactiy of all > > those programs, he would at the very least obtain the most practical > > program on this world. > > I strongly beleive things don't work that way, rather the best way to go > about doing that is to examine those AI programs and collect the best > IDEAS, then create a virtual world like www.creatures.net and apply a > little bit of evolutionary engineering to arive at what truly works. =) You are missing the world's processing capacity. Your ideas are not practicaly implentable (without a great deal of money, which I belive nor you nor me will ever have). > > We take one Altavista, one CYC, one Alice and one NL sentence-builder > > and viola, you've got a very smart, fast and english-speaking system. > > CYC has no value... =P You never know... =] > The Saturn V was used during the Apolo program, 1965-76, it was what put > a man on the moon, and what put an entire space station into orbit > (Skylab) in a single shot. > > In its moon-shot configuration, it stands 13 stories tall, remaining the > tallest launch vehicle in existance. Yes. The effort of blasting a person into outer space somehow cannot compare to the effort of trying to effectively utilize the processing capacity of a planet. Not to meniton AI. > >> AI is AI.... > >> We have used various forms of computational cognition for years, now > >> its time to bring it togeather and achieve human equivalency and > >> beyond! > > > In one program? =] > > You underestimate the power of the dark side of the source. ;) You still belive in a new, revolutionary design that will solve all of your problems? A few years ago I discussed this aspect, toughly beliving in it, with a scienist at the national institute for computerscience. He told me very simply: There is no such thing, whatever has been made to resemble that so far has absolutely no practical value. > >> As does many technological systems... The key is to give it human > >> level conceptualization. > > > Oh, we're talking "levels" again... =] > > Its a yardstick... (Gotta measure things somehow!) Next to the fact that I threw away the idea of intelligence a while ago (it's a been-there-done-that thing), these "yardsticks" appear even more absurd. But say I can tolerate the use of these phrases. > > > What is the deal here anyway? > > > We are obviously working on a program we are trying to sell. I and Neil > > both 'originated' from which, no kidding, has > > something to do with AI devolopment. > > Okay, So there will be money in this eventually... (That increases my > interest). Another good lesson for the promotion guy. =] > > We are not trying to sell you the program, in case you were wondering. > > I meant sell in the other sense, that of trying to get me to accept the > ideas behind your project as being meritous. Yes, that we are trying, which is suspicious not to be that bad at all because in the end you only benerfit. > > Now we sure look silly trying to give you something very usefull for > > free, eh? > > Depends on how useful it turns out to be. ;) Hope you find it at least as usefull as we did. =] C'ya! -- Cellphone: 0038640809676 Don't feel bad about asking/telling me anything, I will always gladly reply. Digging for info? Try AI Meta Search: Http://WWW.AIMetaSearch.Com MesonAI -- If nobody else wants to do it, why shouldn't we?(TM) Http://WWW.MesonAI.Com