[agi] Remembering Caught in the Act
http://www.nytimes.com/2008/09/05/science/05brain.html?_r=3partner=rssnytemc=rssoref=sloginoref=sloginoref=slogin or, indirectly, http://science.slashdot.org/article.pl?sid=08/09/05/0138237from=rss --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] Remembering Caught in the Act
As the article says, this has long been suspected but until now hadn't been demonstrated. Edelman was describing the same phenomena as the remembered present well over a decade ago, and his idea seems to have been loosely inspired by ideas from Freud and James. Remembering seems to be an act of reassembly of stored percepts. If you can intervene and tinker around with the reassembly process you can alter how people remember things, and this seems to be the essence of psychoanalysis. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] Remembering Caught in the Act
On Fri, Sep 5, 2008 at 11:21 AM, Brad Paulsen [EMAIL PROTECTED] wrote: http://www.nytimes.com/2008/09/05/science/05brain.html?_r=3partner=rssnytemc=rssoref=sloginoref=sloginoref=slogin http://www.sciencemag.org/cgi/content/short/1164685 for the original study. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] Remembering Caught in the Act
Er sorry - my question is answered in the interesting Slashdot thread (thanks again): Past studies have shown how many neurons are involved in a single, simple memory. Researchers might be able to isolate a few single neurons in the process of summoning a memory, but that is like saying that they have isolated a few water molecules in the runoff of a giant hydroelectric dam. The practical utility of this is highly questionable. (and much more.. good thread) --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
OK, I'll bite: what's nondeterministic programming if not a contradiction? Again - v. briefly - it's a reality - nondeterministic programming is a reality, so there's no material, mechanistic, software problem in getting a machine to decide either way. The only problem is a logical one of doing it for sensible reasons. And that's the long part - there are a continuous stream of sensible reasons, as there are for current nondeterministic computer choices. Yes, strictly, a nondeterministic *program* can be regarded as a contradiction - i.e. a structured *series* of instructions to decide freely . The way the human mind is programmed is that we are not only free, and have to, *decide* either way about certain decisions, but we are also free to *think* about it - i.e. to decide metacognitively whether and how we decide at all - we continually decide. for example, to put off the decision till later. So the simple reality of being as free to decide and think as you are, is that when you sit down to engage in any task, like write a post, essay, or have a conversation, or almost literally anything, there is no guarantee that you will start, or continue to the 2nd, 3rd, 4th step, let alone complete it. You may jack in your post more or less immediately. This is at once the bane and the blessing of your life, and why you have such extraordinary problems finishing so many things. Procrastination. By contrast, all deterministic/programmed machines and computers are guaranteed to complete any task they begin. (Zero procrastination or deviation). Very different kinds of machines to us. Very different paradigm. (No?) I would say then that the human mind is strictly not so much nondeterministically programmed as briefed. And that's how an AGI will have to function. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] open models, closed models, priors
On Thu, Sep 4, 2008 at 11:17 PM, Abram Demski [EMAIL PROTECTED] wrote: Pei, I sympathize with your care in wording, because I'm very aware of the strange meaning that the word model takes on in formal accounts of semantics. While a cognitive scientist might talk about a person's model of the world, a logician would say that the world is a model of a first-order theory. I do want to avoid the second meaning. But, I don't think I could fare well by saying system instead, because the models are only a part of the larger system... so I'm not sure there is a word that is both neutral and sufficiently meaningful. Yes, the first usage of model is less evil than the second, though it still carry the sense of representing the world as it is and building a one-to-one mapping between the symbols and the objects. As I write in the draft, it is better to take knowledge as a representation of the experience of the system, after summarization and organization. Do you think it is impossible to apply probability to open models/theories/systems, or merely undesirable? Well, to apply probability can be done in many ways. What I have argued (e.g., in http://nars.wang.googlepages.com/wang.bayesianism.pdf) is that if a system is open to new information and works in real time, it is practically impossible to maintain a (consistent) probability distribution among its beliefs --- incremental revision is not supported by the theory, and re-building the distribution from raw data is not affordable. It only works on toy problems and cannot scale up. Pei --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] Remembering Caught in the Act
2008/9/5 Mike Tintner [EMAIL PROTECTED]: Past studies have shown how many neurons are involved in a single, simple memory. Researchers might be able to isolate a few single neurons in the process of summoning a memory, but that is like saying that they have isolated a few water molecules in the runoff of a giant hydroelectric dam. The practical utility of this is highly questionable. (and much more.. good thread) It's true that there isn't much practical utility to be gained from this from the standpoint of designing more intelligent systems. Most people with any interest in neuroscience already thought this to be the case anyway. There's still much more to be learned about how the various stored percepts are reintegrated into a coherent memory or a conscious introspection. It's believed that the thalamus acts as a kind of central switching hub dynamically recruiting neural assemblies from diverse areas of the cortex, which Edelman called the dynamic core, although the exact details of how this works remains to be characterised. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
2008/9/5 Mike Tintner [EMAIL PROTECTED]: By contrast, all deterministic/programmed machines and computers are guaranteed to complete any task they begin. If only such could be guaranteed! We would never have system hangs, dead locks. Even if it could be made so, computer systems would not always want to do so. Have you every had a programmed computer system say to you. This program is not responding, do you wish to terminate it. There is no reason in principle why the decision to terminate the program couldn't be made automatically. (Zero procrastination or deviation). Multi-tasking systems deviate all the time... Very different kinds of machines to us. Very different paradigm. (No?) We commonly talk about single program systems because they are generally interesting, and can be analysed simply. My discussion on self-modifying systems ignored the interrupt driven multi-tasking nature of the system I want to build, because that makes analysis a lot more hard. I will still be building an interrupt driven, multi tasking system. Will Pearson --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
MT:By contrast, all deterministic/programmed machines and computers are guaranteed to complete any task they begin. Will:If only such could be guaranteed! We would never have system hangs, dead locks. Even if it could be made so, computer systems would not always want to do so. Will, That's a legalistic, not a valid objection, (although heartfelt!).In the above case, the computer is guaranteed to hang - and it does, strictly, complete its task. What's happened is that you have had imperfect knowledge of the program's operations. Had you known more, you would have known that it would hang. Were your computer like a human mind, it would have been able to say (as you/we all do) - well if that part of the problem is going to be difficult, I'll ignore it or.. I'll just make up an answer... or by God I'll keep trying other ways until I do solve this.. or... .. or ... Computers, currently, aren't free thinkers. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Language modeling (was Re: [agi] draft for comment)
--- On Thu, 9/4/08, Pei Wang [EMAIL PROTECTED] wrote: I guess you still see NARS as using model-theoretic semantics, so you call it symbolic and contrast it with system with sensors. This is not correct --- see http://nars.wang.googlepages.com/wang.semantics.pdf and http://nars.wang.googlepages.com/wang.AI_Misconceptions.pdf I mean NARS is symbolic in the sense that you write statements in Narsese like raven - bird 0.97, 0.92 (probability=0.97, confidence=0.92). I realize that the meanings of raven and bird are determined by their relations to other symbols in the knowledge base and that the probability and confidence change with experience. But in practice you are still going to write statements like this because it is the easiest way to build the knowledge base. You aren't going to specify the brightness of millions of pixels in a vision system in Narsese, and there is no mechanism I am aware of to collect this knowledge from a natural language text corpus. There is no mechanism to add new symbols to the knowledge base through experience. You have to explicitly add them. You have made this point on CPU power several times, and I'm still not convinced that the bottleneck of AI is hardware capacity. Also, there is no reason to believe an AGI must be designed in a biologically plausible way. Natural language has evolved to be learnable on a massively parallel network of slow computing elements. This should be apparent when we compare successful language models with unsuccessful ones. Artificial language models usually consist of tokenization, parsing, and semantic analysis phases. This does not work on natural language because artificial languages have precise specifications and natural languages do not. No two humans use exactly the same language, nor does the same human at two points in time. Rather, language is learnable by example, so that each message causes the language of the receiver to be a little more like that of the sender. Children learn semantics before syntax, which is the opposite order from which you would write an artificial language interpreter. An example of a successful language model is a search engine. We know that most of the meaning of a text document depends only on the words it contains, ignoring word order. A search engine matches the semantics of the query with the semantics of a document mostly by matching words, but also by matching semantically related words like water to wet. Here is an example of a computationally intensive but biologically plausible language model. A semantic model is a word-word matrix A such that A_ij is the degree to which words i and j are related, which you can think of as the probability of finding i and j together in a sliding window over a huge text corpus. However, semantic relatedness is a fuzzy identity relation, meaning it is reflexive, commutative, and transitive. If i is related to j and j to k, then i is related to k. Deriving transitive relations in A, also known as latent semantic analysis, is performed by singular value decomposition, factoring A = USV where S is diagonal, then discarding the small terms of S, which has the effect of lossy compression. Typically, A has about 10^6 elements and we keep only a few hundred elements of S. Fortunately there is a parallel algorithm that incrementally updates the matrices as the system learns: a 3 layer neural network where S is the hidden layer (which can grow) and U and V are weight matrices. [1]. Traditional language processing has failed because the task of converting natural language statements like ravens are birds to formal language is itself an AI problem. It requires humans who have already learned what ravens are and how to form and recognize grammatically correct sentences so they understand all of the hundreds of ways to express the same statement. You have to have human level understand of the logic to realize that ravens are coming doesn't mean ravens - coming. If you solve the translation problem, then you must have already solved the natural language problem. You can't take a shortcut directly to the knowledge base, tempting as it might be. You have to learn the language first, going through all the childhood stages. I would have hoped we have learned a lesson from Cyc. 1. Gorrell, Genevieve (2006), Generalized Hebbian Algorithm for Incremental Singular Value Decomposition in Natural Language Processing, Proceedings of EACL 2006, Trento, Italy. http://www.aclweb.org/anthology-new/E/E06/E06-1013.pdf -- Matt Mahoney, [EMAIL PROTECTED] --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: Language modeling (was Re: [agi] draft for comment)
On Fri, Sep 5, 2008 at 11:15 AM, Matt Mahoney [EMAIL PROTECTED] wrote: --- On Thu, 9/4/08, Pei Wang [EMAIL PROTECTED] wrote: I guess you still see NARS as using model-theoretic semantics, so you call it symbolic and contrast it with system with sensors. This is not correct --- see http://nars.wang.googlepages.com/wang.semantics.pdf and http://nars.wang.googlepages.com/wang.AI_Misconceptions.pdf I mean NARS is symbolic in the sense that you write statements in Narsese like raven - bird 0.97, 0.92 (probability=0.97, confidence=0.92). I realize that the meanings of raven and bird are determined by their relations to other symbols in the knowledge base and that the probability and confidence change with experience. But in practice you are still going to write statements like this because it is the easiest way to build the knowledge base. Yes. You aren't going to specify the brightness of millions of pixels in a vision system in Narsese, and there is no mechanism I am aware of to collect this knowledge from a natural language text corpus. Of course not. To have visual experience, there must be a devise to convert visual signals into internal representation in Narsese. I never suggested otherwise. There is no mechanism to add new symbols to the knowledge base through experience. You have to explicitly add them. New symbols either come from the outside in experience (experience can be verbal), or composed by the concept-formation rules from existing ones. The latter case is explained in my book. Natural language has evolved to be learnable on a massively parallel network of slow computing elements. This should be apparent when we compare successful language models with unsuccessful ones. Artificial language models usually consist of tokenization, parsing, and semantic analysis phases. This does not work on natural language because artificial languages have precise specifications and natural languages do not. It depends on which aspect of the language you talk about. Narsese has precise specifications in syntax, but the meaning of the terms is a function of experience, and change from time to time. No two humans use exactly the same language, nor does the same human at two points in time. Rather, language is learnable by example, so that each message causes the language of the receiver to be a little more like that of the sender. Same thing in NARS --- if two implementations of NARS have different experience, they will disagree on what is the meaning of a term. When they begin to learn natural language, it will also be true for grammar. Since I haven't done any concrete NLP yet, I don't expect you to believe me on the second point, but you cannot rule out that possibility just because no traditional system can do that. Children learn semantics before syntax, which is the opposite order from which you would write an artificial language interpreter. NARS indeed can learn semantics before syntax --- see http://nars.wang.googlepages.com/wang.roadmap.pdf I won't comment on the following detailed statements, since I agree with your criticism on the traditional processing of formal language, but that is not how NARS handles languages. Don't think NARS as another Cyc just because both use formal language. The same ravens are birds in these two systems are treated very differently in them. Pei An example of a successful language model is a search engine. We know that most of the meaning of a text document depends only on the words it contains, ignoring word order. A search engine matches the semantics of the query with the semantics of a document mostly by matching words, but also by matching semantically related words like water to wet. Here is an example of a computationally intensive but biologically plausible language model. A semantic model is a word-word matrix A such that A_ij is the degree to which words i and j are related, which you can think of as the probability of finding i and j together in a sliding window over a huge text corpus. However, semantic relatedness is a fuzzy identity relation, meaning it is reflexive, commutative, and transitive. If i is related to j and j to k, then i is related to k. Deriving transitive relations in A, also known as latent semantic analysis, is performed by singular value decomposition, factoring A = USV where S is diagonal, then discarding the small terms of S, which has the effect of lossy compression. Typically, A has about 10^6 elements and we keep only a few hundred elements of S. Fortunately there is a parallel algorithm that incrementally updates the matrices as the system learns: a 3 layer neural network where S is the hidden layer (which can grow) and U and V are weight matrices. [1]. Traditional language processing has failed because the task of converting natural language statements like ravens are birds to formal language is itself an AI problem. It
Re: Real vs. simulated environments (was Re: [agi] draft for comment.. P.S.)
Matt, FINALLY, someone here is saying some of the same things that I have been saying. With general agreement with your posting, I will make some comments... On 9/4/08, Matt Mahoney [EMAIL PROTECTED] wrote: --- On Thu, 9/4/08, Valentina Poletti [EMAIL PROTECTED] wrote: Ppl like Ben argue that the concept/engineering aspect of intelligence is independent of the type of environment. That is, given you understand how to make it in a virtual environment you can then tarnspose that concept into a real environment more safely. This is probably a good starting point, to avoid beating the world up during the debugging process. Some other ppl on the other hand believe intelligence is a property of humans only. Only people who haven't had a pet believe such things. I have seen too many animals find clever solutions to problems. So you have to simulate every detail about humans to get that intelligence. I'd say that among the two approaches the first one (Ben's) is safer and more realistic. The issue is not what is intelligence, but what do you want to create? In order for machines to do more work for us, they may need language and vision, which we associate with human intelligence. Not necessarily, as even text-interfaced knowledge engines can handily outperform humans in many complex problem solving tasks. The still open question is: What would best do what we need done but can NOT presently do (given computers, machinery, etc.). So far, the talk here on this forum has been about what we could do and how we might do it, rather than about what we NEED done. Right now, we NEED resources to work productively in the directions that we have been discussing, yet the combined intelligence of those here on this forum is apparently unable to solve even this seemingly trivial problem. Perhaps something more than raw intelligence is needed? But building artificial humans is not necessarily useful. We already know how to create humans, and we are doing so at an unsustainable rate. I suggest that instead of the imitation game (Turing test) for AI, we should use a preference test. If you prefer to talk to a machine vs. a human, then the machine passes the test. YES, like what is it that our AGI can do that we need done but can NOT presently do? Prediction is central to intelligence. If you can predict a text stream, then for any question Q and any answer A, you can compute the probability distribution P(A|Q) = P(QA)/P(Q). This passes the Turing test. More importantly, it allows you to output max_A P(QA), the most likely answer from a group of humans. This passes the preference test because a group is usually more accurate than any individual member. (It may fail a Turing test for giving too few wrong answers, a problem Turing was aware of in 1950 when he gave an example of a computer incorrectly answering an arithmetic problem). Unfortunately, this also tests the ability to incorporate the very misunderstandings that presently limit our thinking. We need to give credit for compression algorithms that cleans up our grammar, corrects our technical errors, etc., as these can probably be done in the process of better compressing the text. Text compression is equivalent to AI because we have already solved the coding problem. Given P(x) for string x, we know how to optimally and efficiently code x in log_2(1/P(x)) bits (e.g. arithmetic coding). Text compression has an advantage over the Turing or preference tests in that that incremental progress in modeling can be measured precisely and the test is repeatable and verifiable. If I want to test a text compressor, it is important to use real data (human generated text) rather than simulated data, i.e. text generated by a program. Otherwise, I know there is a concise code for the input data, which is the program that generated it. When you don't understand the source distribution (i.e. the human brain), the problem is much harder, and you have a legitimate test. Wouldn't it be better to understand the problem domain while ignoring human (mis)understandings? After all, if humans need an AGI to work in a difficult domain, it is probably made more difficult by incorporating human misunderstandings. Of course, humans state human problems, so it is important to be able to semantically communicate, but also useful to separate the communications from the problems. I understand that Ben is developing AI for virtual worlds. This might produce interesting results, but I wouldn't call it AGI. The value of AGI is on the order of US $1 quadrillion. It is a global economic system running on a smarter internet. I believe that any attempt to develop AGI on a budget of $1 million or $1 billion or $1 trillion is just wishful thinking. I think that a billion or so, divided up into small pieces to fund EVERY disparate approach to see where the low hanging fruit is, would go a LONG way in guiding subsequent billions. I doubt that it would
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
Mike, Will's objection is not quite so easily dismissed. You need to argue that there is an alternative, not just that Will's is more of the same. --Abram On Fri, Sep 5, 2008 at 9:34 AM, Mike Tintner [EMAIL PROTECTED] wrote: MT:By contrast, all deterministic/programmed machines and computers are guaranteed to complete any task they begin. Will:If only such could be guaranteed! We would never have system hangs, dead locks. Even if it could be made so, computer systems would not always want to do so. Will, That's a legalistic, not a valid objection, (although heartfelt!).In the above case, the computer is guaranteed to hang - and it does, strictly, complete its task. What's happened is that you have had imperfect knowledge of the program's operations. Had you known more, you would have known that it would hang. Were your computer like a human mind, it would have been able to say (as you/we all do) - well if that part of the problem is going to be difficult, I'll ignore it or.. I'll just make up an answer... or by God I'll keep trying other ways until I do solve this.. or... .. or ... Computers, currently, aren't free thinkers. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
Mike, The philosophical paradigm I'm assuming is that the only two alternatives are deterministic and random. Either the next state is completely determined by the last, or it is only probabilistically determined. Deterministic does not mean computable, since physical processes can be totally well-defined without being computable (take Newton's physics for example). So, 1) Is the next action that your creativity machine will take intended to be uniquely defined, given its experience and inputs? 2) is the next action intended to be computable from the experience and inputs? Meaning (approximately), could the creativity machine be implemented on a computer? --Abram On Fri, Sep 5, 2008 at 9:26 AM, Mike Tintner [EMAIL PROTECTED] wrote: Abram: In that case I do not see how your view differs from simplistic dualism, as Terren cautioned. If your goal is to make a creativity machine, in what sense would the machine be non-algorithmic? Physical random processes? Abram, You're operating within a philosophical paradigm that says all actions and problemsolving must be preprogrammed. Nothing else is possible. That ignores the majority of real life problems where no program is possible, period. Sometimes the best plan is no plan If you're confronted with the task of finding something in a foreign territory, you simply don't (and couldn't) have the luxury of a program. All you have is a rough idea, as opposed to an algorithm, of the sort of things you can do. You know roughly what you're looking for - an object somewhere in that territory. You know roughly how to travel and put one foor in front of the other and avoid obstacles and pick things up etc. (Let's say - you have to find a key that has been lost somewhere in a house). Well you certainly don't have an algorithm for finding a lost key in a house. In fact, if you or anyone would care to spend 5 mins on this problem, you would start to realise that no algorithm is possible. Check out Kauffman's interview on edge.com. for similar problems arguments . So what do/can you do? Make it up as you go along. Start somewhere and keep going, and after a while if that doesn't work, try somewhere and something else... But there's no algorithm for this. Just as there is, or was, no algorithm for your putting the pieces of a jigsaw puzzle together (a much simpler, more tightly defined problem). You just got stuck in. Somewhere. Anywhere reasonable. Algorithms, from a human POV, are for literal people who have to do things by the book - people with a compulsive obsessional disorder - who can't bear to confront a blank page. :).V useful *after* you've solved a problem, but not in the beginning There are no physical, computational, mechanical reasons why machines can't be designed on these principles - to proceed with rough ideas of what to do, freely consulting and combining options and looking around for fresh ones, as they go along, rather than following a preprogrammed list. P.S. Nothing in this is strictly random - as in a narrow AI, randomly, blindly. working its way through a preprogrammed list. You only try options that are appropriate - routes that appear likely to lead to your goal. I would call this unstructured but not (blindly) random thinking. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] What is Friendly AI?
Vladamir, On 9/4/08, Vladimir Nesov [EMAIL PROTECTED] wrote: On Thu, Sep 4, 2008 at 12:02 PM, Valentina Poletti [EMAIL PROTECTED] wrote: Also, Steve made another good point here: loads of people at any moment do whatever they can to block the advancement and progress of human beings as it is now. How will those people react to a progress as advanced as AGI? That's why I keep stressing the social factor in intelligence as very important part to consider. No, it's not important, unless these people start to pose a serious threat to the project. Here we are, lunch-money funded, working on the project with the MOST economic potential of any project in the history of man. NO ONE will invest the few millions needed to check out the low hanging fruit and kick this thing into high gear. Sure, no one is holding guns to investors' heads and saying don't invest, but neither is it socially acceptable to invest in such directions. That social system is crafted by the Christian majority here in the U.S. Hence, I see U.S. Christians as being a THE really SERIOUS threat to AGI. You need to care about what is the correct answer, not what is a popular one, in the case where popular answer is dictated by ignorance. As Reverse Reductio ad Absurdum shows ever so well, you can't even understand the answers without some education. This is akin to learning that a Game Theory solution consists of a list of probabilities, with the final decision being made as a weighted random decision. Hence, there appears to be NO prospect of an AGI being useful to people who lack this sort of education, as nearly all of the population and all of the world leaders now lack. Given a ubiquitous understanding of these principles, people are probably smart enough to figure things out for themselves, so AGIs may not even be needed. Most disputes are NOT about what is the best answer, but rather about what the goal is. Special methods like Reverse Reductio ad Absurdum are needed in situations with conflicting goals. The Koran states that most evil is done by people who think they are doing good. However, Christians, seeing another competing religious book as itself being evil, reject all of the wisdom therein, and in their misdirected actions, confirm this very statement. When the majority of people reject wisdom simply because of its source, and AGIs must necessarily displace religions as they identify the misstatements made therein, it seems pretty obvious to me that a war without limits lies ahead between the Christian majority and AGIs. Steve Richfield --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
Abram, I don't understand why.how I need to argue an alternative - please explain. If it helps, a deterministic, programmed machine can, at any given point, only follow one route through a given territory or problem space or maze - even if surprising *appearing* to halt/deviate from the plan - to the original, less-than-omniscient-of-what-he-hath-wrought programmer. (A fundamental programming problem, right?) A creative free machine, like a human, really can follow any of what may be a vast range of routes - and you really can't predict what it will do or, at a basic level, be surprised by it. Mike, Will's objection is not quite so easily dismissed. You need to argue that there is an alternative, not just that Will's is more of the same. --Abram On Fri, Sep 5, 2008 at 9:34 AM, Mike Tintner [EMAIL PROTECTED] wrote: MT:By contrast, all deterministic/programmed machines and computers are guaranteed to complete any task they begin. Will:If only such could be guaranteed! We would never have system hangs, dead locks. Even if it could be made so, computer systems would not always want to do so. Will, That's a legalistic, not a valid objection, (although heartfelt!).In the above case, the computer is guaranteed to hang - and it does, strictly, complete its task. What's happened is that you have had imperfect knowledge of the program's operations. Had you known more, you would have known that it would hang. Were your computer like a human mind, it would have been able to say (as you/we all do) - well if that part of the problem is going to be difficult, I'll ignore it or.. I'll just make up an answer... or by God I'll keep trying other ways until I do solve this.. or... .. or ... Computers, currently, aren't free thinkers. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
Hi Mike, comments below... --- On Fri, 9/5/08, Mike Tintner [EMAIL PROTECTED] wrote: Again - v. briefly - it's a reality - nondeterministic programming is a reality, so there's no material, mechanistic, software problem in getting a machine to decide either way. This is inherently dualistic to say this. On one hand you're calling it a 'reality' and on the other you're denying the influence of material or mechanism. What exactly is deciding then, a soul? How do you get one of those into an AI? Yes, strictly, a nondeterministic *program* can be regarded as a contradiction - i.e. a structured *series* of instructions to decide freely. At some point you will have to explain how this deciding freely works. As of now, all you have done is name it. The way the human mind is programmed is that we are not only free, and have to, *decide* either way about certain decisions, but we are also free to *think* about it - i.e. to decide metacognitively whether and how we decide at all - we continually decide. for example, to put off the decision till later. There is an entire school of thought, quite mainstream now, in cognitive science that says that what appears to be free will is an illusion. Of course, you can say that you are free to choose whatever you like, but that only speaks to the strength of the illusion - that in itself is not enough to disprove the claim. In fact, it is plain to see that if you do not commit yourself to this view (free will as illusion), you are either a dualist, or you must invoke some kind of probabilistic mechanism (as some like Penrose have done by saying that the free-will buck stops at the level of quantum mechanics). So, Mike, is free will: 1) an illusion based on some kind of unpredictable, complex but *deterministic* interaction of physical components 2) the result of probabilistic physics - a *non-deterministic* interaction described by something like quantum mechanics 3) the expression of our god-given spirit, or some other non-physical mover of physical things By contrast, all deterministic/programmed machines and computers are guaranteed to complete any task they begin. (Zero procrastination or deviation). Very different kinds of machines to us. Very different paradigm. (No?) I think the difference of paradigm between computers and humans is not that one is deterministic and one isn't, but rather that one is a paradigm of top-down, serialized control, and the other is bottom-up, massively parallel, and emergent. It comes down to design vs. emergence. Terren --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
Mike, On Fri, Sep 5, 2008 at 1:15 PM, Mike Tintner [EMAIL PROTECTED] wrote: Abram, I don't understand why.how I need to argue an alternative - please explain. I am not sure what to say, but here is my view of the situation. You are claiming that there is a broad range of things that algorithmic systems cannot do. You gave some examples. William took a couple of these examples and argued that they are routinely done by multi-tasking systems. You say that those methods do not really count, because they reduce to normal computation. I say that that is not a valid response, because that was exactly Will's point, that they do reduce to normal computations. To make your objection work, you need to argue that humans do not do the same sort of thing when we change our minds about something. If it helps, a deterministic, programmed machine can, at any given point, only follow one route through a given territory or problem space or maze - even if surprising *appearing* to halt/deviate from the plan - to the original, less-than-omniscient-of-what-he-hath-wrought programmer. (A fundamental programming problem, right?) A creative free machine, like a human, really can follow any of what may be a vast range of routes - and you really can't predict what it will do or, at a basic level, be surprised by it. It still sounds like you are describing physical randomness. --Abram --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: Language modeling (was Re: [agi] draft for comment)
--- On Fri, 9/5/08, Pei Wang [EMAIL PROTECTED] wrote: NARS indeed can learn semantics before syntax --- see http://nars.wang.googlepages.com/wang.roadmap.pdf Yes, I see this corrects many of the problems with Cyc and with traditional language models. I didn't see a description of a mechanism for learning new terms in your other paper. Clearly this could be added, although I believe it should be a statistical process. I am interested in determining the computational cost of language modeling. The evidence I have so far is that it is high. I believe the algorithmic complexity of a model is 10^9 bits. This is consistent with Turing's 1950 prediction that AI would require this much memory, with Landauer's estimate of human long term memory, and is about how much language a person processes by adulthood assuming an information content of 1 bit per character as Shannon estimated in 1950. This is why I use a 1 GB data set in my compression benchmark. However there is a 3 way tradeoff between CPU speed, memory, and model accuracy (as measured by compression ratio). I added two graphs to my benchmark at http://cs.fit.edu/~mmahoney/compression/text.html (below the main table) which shows this clearly. In particular the size-memory tradeoff is an almost perfectly straight line (with memory on a log scale) over tests of 104 compressors. These tests suggest to me that CPU and memory are indeed bottlenecks to language modeling. The best models in my tests use simple semantic and grammatical models, well below adult human level. The 3 top programs on the memory graph map words to tokens using dictionaries that group semantically and syntactically related words together, but only one (paq8hp12any) uses a semantic space of more than one dimension. All have large vocabularies, although not implausibly large for an educated person. Other top programs like nanozipltcb and WinRK use smaller dictionaries and strictly lexical models. Lesser programs model only at the n-gram level. I don't yet have an answer to my question, but I believe efficient human-level NLP will require hundreds of GB or perhaps 1 TB of memory. The slowest programs are already faster than real time, given that equivalent learning in humans would take over a decade. I think you could use existing hardware in a speed-memory tradeoff to get real time NLP, but it would not be practical for doing experiments where each source code change requires training the model from scratch. Model development typically requires thousands of tests. -- Matt Mahoney, [EMAIL PROTECTED] --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: Language modeling (was Re: [agi] draft for comment)
On Fri, Sep 5, 2008 at 6:15 PM, Matt Mahoney [EMAIL PROTECTED] wrote: --- On Fri, 9/5/08, Pei Wang [EMAIL PROTECTED] wrote: NARS indeed can learn semantics before syntax --- see http://nars.wang.googlepages.com/wang.roadmap.pdf Yes, I see this corrects many of the problems with Cyc and with traditional language models. I didn't see a description of a mechanism for learning new terms in your other paper. Clearly this could be added, although I believe it should be a statistical process. I don't have a separate paper on term composition, so you'd have to read my book. It is indeed a statistical process, in the sense that most of the composed terms won't be useful, so will be forgot gradually. Only the useful patterns will be kept for long time in the form of compound terms. I am interested in determining the computational cost of language modeling. The evidence I have so far is that it is high. I believe the algorithmic complexity of a model is 10^9 bits. This is consistent with Turing's 1950 prediction that AI would require this much memory, with Landauer's estimate of human long term memory, and is about how much language a person processes by adulthood assuming an information content of 1 bit per character as Shannon estimated in 1950. This is why I use a 1 GB data set in my compression benchmark. I see your point, though I think to analyze this problem in terms of computational complexity is not the correct way to go, because this process does not follow a predetermined algorithm. Instead, language learning is an incremental process, without a well-defined beginning and ending. However there is a 3 way tradeoff between CPU speed, memory, and model accuracy (as measured by compression ratio). I added two graphs to my benchmark at http://cs.fit.edu/~mmahoney/compression/text.html (below the main table) which shows this clearly. In particular the size-memory tradeoff is an almost perfectly straight line (with memory on a log scale) over tests of 104 compressors. These tests suggest to me that CPU and memory are indeed bottlenecks to language modeling. The best models in my tests use simple semantic and grammatical models, well below adult human level. The 3 top programs on the memory graph map words to tokens using dictionaries that group semantically and syntactically related words together, but only one (paq8hp12any) uses a semantic space of more than one dimension. All have large vocabularies, although not implausibly large for an educated person. Other top programs like nanozipltcb and WinRK use smaller dictionaries and strictly lexical models. Lesser programs model only at the n-gram level. Like to many existing AI works, my disagreement with you is not that much on the solution you proposed (I can see the value), but on the problem you specified as the goal of AI. For example, I have no doubt about the theoretical and practical values of compression, but don't think it has much to do with intelligence. I don't think this kind of issue can be efficient handled by email discussion like this one. I've been thinking about to write a paper to compare my ideas with the ideas represented by AIXI, which is closely related to yours, though this project hasn't got enough priority in my to-do list. Hopefully I'll find the time to make myself clear on this topic. I don't yet have an answer to my question, but I believe efficient human-level NLP will require hundreds of GB or perhaps 1 TB of memory. The slowest programs are already faster than real time, given that equivalent learning in humans would take over a decade. I think you could use existing hardware in a speed-memory tradeoff to get real time NLP, but it would not be practical for doing experiments where each source code change requires training the model from scratch. Model development typically requires thousands of tests. I guess we are exploring very different paths in NLP, and now it is too early to tell which one will do better. Pei --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: Language modeling (was Re: [agi] draft for comment)
--- On Fri, 9/5/08, Pei Wang [EMAIL PROTECTED] wrote: Like to many existing AI works, my disagreement with you is not that much on the solution you proposed (I can see the value), but on the problem you specified as the goal of AI. For example, I have no doubt about the theoretical and practical values of compression, but don't think it has much to do with intelligence. In http://cs.fit.edu/~mmahoney/compression/rationale.html I explain why text compression is an AI problem. To summarize, if you know the probability distribution of text, then you can compute P(A|Q) for any question Q and answer A to pass the Turing test. Compression allows you to precisely measure the accuracy of your estimate of P. Compression (actually, word perplexity) has been used since the early 1990's to measure the quality of language models for speech recognition, since it correlates well with word error rate. The purpose of this work is not to solve general intelligence, such as the universal intelligence proposed by Legg and Hutter [1]. That is not computable, so you have to make some arbitrary choice with regard to test environments about what problems you are going to solve. I believe the goal of AGI should be to do useful work for humans, so I am making a not so arbitrary choice to solve a problem that is central to what most people regard as useful intelligence. I had hoped that my work would lead to an elegant theory of AI, but that hasn't been the case. Rather, the best compression programs were developed as a series of thousands of hacks and tweaks, e.g. change a 4 to a 5 because it gives 0.002% better compression on the benchmark. The result is an opaque mess. I guess I should have seen it coming, since it is predicted by information theory (e.g. [2]). Nevertheless the architectures of the best text compressors are consistent with cognitive development models, i.e. phoneme (or letter) sequences - lexical - semantics - syntax, which are themselves consistent with layered neural architectures. I already described a neural semantic model in my last post. I also did work supporting Hutchens and Alder showing that lexical models can be learned from n-gram statistics, consistent with the observation that babies learn the rules for segmenting continuous speech before they learn any words [3]. I agree it should also be clear that semantics is learned before grammar, contrary to the way artificial languages are processed. Grammar requires semantics, but not the other way around. Search engines work using semantics only. Yet we cannot parse sentences like I ate pizza with Bob, I ate pizza with pepperoni, I ate pizza with chopsticks, without semantics. My benchmark does not prove that there aren't better language models, but it is strong evidence. It represents the work of about 100 researchers who have tried and failed to find more accurate, faster, or less memory intensive models. The resource requirements seem to increase as we go up the chain from n-grams to grammar, contrary to symbolic approaches. This is my argument why I think AI is bound by lack of hardware, not lack of theory. 1. Legg, Shane, and Marcus Hutter (2006), A Formal Measure of Machine Intelligence, Proc. Annual machine learning conference of Belgium and The Netherlands (Benelearn-2006). Ghent, 2006. http://www.vetta.org/documents/ui_benelearn.pdf 2. Legg, Shane, (2006), Is There an Elegant Universal Theory of Prediction?, Technical Report IDSIA-12-06, IDSIA / USI-SUPSI, Dalle Molle Institute for Artificial Intelligence, Galleria 2, 6928 Manno, Switzerland. http://www.vetta.org/documents/IDSIA-12-06-1.pdf 3. M. Mahoney (2000), A Note on Lexical Acquisition in Text without Spaces, http://cs.fit.edu/~mmahoney/dissertation/lex1.html -- Matt Mahoney, [EMAIL PROTECTED] --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
AI isn't cheap (was Re: Real vs. simulated environments (was Re: [agi] draft for comment.. P.S.))
--- On Fri, 9/5/08, Steve Richfield [EMAIL PROTECTED] wrote: I think that a billion or so, divided up into small pieces to fund EVERY disparate approach to see where the low hanging fruit is, would go a LONG way in guiding subsequent billions. I doubt that it would take a trillion to succeed. Sorry, the low hanging fruit was all picked by the early 1960's. By then we had neural networks [1,6,7,11,12], natural language processing and language translation [2], models of human decision making [3], automatic theorem proving [4,8,10], natural language databases [5], game playing programs [9,13], optical character recognition [14], handwriting and speech recognition [15], and important theoretical work [16,17,18]. Since then we have had mostly just incremental improvements. Big companies like Google and Microsoft have strong incentives to develop AI and have billions to spend. Maybe the problem really is hard. References 1. Ashby, W. Ross (1960), Design for a Brain, 2’nd Ed., London: Wiley. Describes a 4 neuron electromechanical neural network. 2. Borko, Harold (1967), Automated Language Processing, The State of the Art, New York: Wiley. Cites 72 NLP systems prior to 1965, and the 1959-61 U.S. government Russian-English translation project. 3. Feldman, Julian (1961), Simulation of Behavior in the Binary Choice Experiment, Proceedings of the Western Joint Computer Conference 19:133-144 4. Gelernter, H. (1959), Realization of a Geometry-Theorem Proving Machine, Proceedings of an International Conference on Information Processing, Paris: UNESCO House, pp. 273-282. 5. Green, Bert F. Jr., Alice K. Wolf, Carol Chomsky, and Kenneth Laughery (1961), Baseball: An Automatic Question Answerer, Proceedings of the Western Joint Computer Conference, 19:219-224. 6. Hebb, D. O. (1949), The Organization of Behavior, New York: Wiley. Proposed the first model of learning in neurons: when two neurons fire simultaneously, the synapse between them becomes stimulating. 7. McCulloch, Warren S., and Walter Pitts (1943), A logical calculus of the ideas immanent in nervous activity, Buletin of Mathematical Biophysics (5) pp. 115-133. 8. Newell, Allen, J. C. Shaw, H. A. Simon (1957), Empirical Explorations with the Logic Theory Machine: A Case Study in Heuristics, Proceedings of the Western Joint Computer Conference, 15:218-239. 9. Newell, Allen, J. C. Shaw, and H. A. Simon (1958), Chess-Playing Programs and the Problem of Complexity, IBM Journal of Research and Development, 2:320-335. 10. Newell, Allen, H. A. Simon (1961), GPS: A Program that Simulates Human Thought, Lernende Automaten, Munich: R. Oldenbourg KG. 11. Rochester, N., J. J. Holland, L. H. Haibt, and Wl L. Duda (1956), Tests on a cell assembly theory of the action of the brain, using a large digital computer, IRE Transactions on Information Theory IT-2: pp. 80-93. 12. Rosenblatt, F. (1958), The perceptron: a probabilistic model for information storage and organization in the brain, Psychological Review (65) pp. 386-408. 13. Samuel, A. L. (1959), Some Studies in Machine Learning using the Game of Checkers, IBM Journal of Research and Development, 3:211-229. 14. Selfridge, Oliver G., Ulric Neisser (1960), Pattern Recognition by Machine, Scientific American, Aug., 203:60-68. 15. Uhr, Leonard, Charles Vossler (1963) A Pattern-Recognition Program that Generates, Evaluates, and Adjusts its own Operators, Computers and Thought, E. A. Feigenbaum and J. Feldman eds, New York: McGraw Hill, pp. 251-268. 16. Turing, A. M., (1950) Computing Machinery and Intelligence, Mind, 59:433-460. 17. Shannon, Claude, and Warren Weaver (1949), The Mathematical Theory of Communication, Urbana: University of Illinois Press. 18. Minsky, Marvin (1961), Steps toward Artificial Intelligence, Proceedings of the Institute of Radio Engineers, 49:8-30. -- Matt Mahoney, [EMAIL PROTECTED] --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: Language modeling (was Re: [agi] draft for comment)
Matt, Thanks for taking the time to explain your ideas in detail. As I said, our different opinions on how to do AI come from our very different understanding of intelligence. I don't take passing Turing Test as my research goal (as explained in http://nars.wang.googlepages.com/wang.logic_intelligence.pdf and http://nars.wang.googlepages.com/wang.AI_Definitions.pdf). I disagree with Hutter's approach, not because his SOLUTION is not computable, but because his PROBLEM is too idealized and simplified to be relevant to the actual problems of AI. Even so, I'm glad that we can still agree on somethings, like semantics comes before syntax. In my plan for NLP, there won't be separate 'parsing' and 'semantic mapping' stages. I'll say more when I have concrete results to share. Pei On Fri, Sep 5, 2008 at 8:39 PM, Matt Mahoney [EMAIL PROTECTED] wrote: --- On Fri, 9/5/08, Pei Wang [EMAIL PROTECTED] wrote: Like to many existing AI works, my disagreement with you is not that much on the solution you proposed (I can see the value), but on the problem you specified as the goal of AI. For example, I have no doubt about the theoretical and practical values of compression, but don't think it has much to do with intelligence. In http://cs.fit.edu/~mmahoney/compression/rationale.html I explain why text compression is an AI problem. To summarize, if you know the probability distribution of text, then you can compute P(A|Q) for any question Q and answer A to pass the Turing test. Compression allows you to precisely measure the accuracy of your estimate of P. Compression (actually, word perplexity) has been used since the early 1990's to measure the quality of language models for speech recognition, since it correlates well with word error rate. The purpose of this work is not to solve general intelligence, such as the universal intelligence proposed by Legg and Hutter [1]. That is not computable, so you have to make some arbitrary choice with regard to test environments about what problems you are going to solve. I believe the goal of AGI should be to do useful work for humans, so I am making a not so arbitrary choice to solve a problem that is central to what most people regard as useful intelligence. I had hoped that my work would lead to an elegant theory of AI, but that hasn't been the case. Rather, the best compression programs were developed as a series of thousands of hacks and tweaks, e.g. change a 4 to a 5 because it gives 0.002% better compression on the benchmark. The result is an opaque mess. I guess I should have seen it coming, since it is predicted by information theory (e.g. [2]). Nevertheless the architectures of the best text compressors are consistent with cognitive development models, i.e. phoneme (or letter) sequences - lexical - semantics - syntax, which are themselves consistent with layered neural architectures. I already described a neural semantic model in my last post. I also did work supporting Hutchens and Alder showing that lexical models can be learned from n-gram statistics, consistent with the observation that babies learn the rules for segmenting continuous speech before they learn any words [3]. I agree it should also be clear that semantics is learned before grammar, contrary to the way artificial languages are processed. Grammar requires semantics, but not the other way around. Search engines work using semantics only. Yet we cannot parse sentences like I ate pizza with Bob, I ate pizza with pepperoni, I ate pizza with chopsticks, without semantics. My benchmark does not prove that there aren't better language models, but it is strong evidence. It represents the work of about 100 researchers who have tried and failed to find more accurate, faster, or less memory intensive models. The resource requirements seem to increase as we go up the chain from n-grams to grammar, contrary to symbolic approaches. This is my argument why I think AI is bound by lack of hardware, not lack of theory. 1. Legg, Shane, and Marcus Hutter (2006), A Formal Measure of Machine Intelligence, Proc. Annual machine learning conference of Belgium and The Netherlands (Benelearn-2006). Ghent, 2006. http://www.vetta.org/documents/ui_benelearn.pdf 2. Legg, Shane, (2006), Is There an Elegant Universal Theory of Prediction?, Technical Report IDSIA-12-06, IDSIA / USI-SUPSI, Dalle Molle Institute for Artificial Intelligence, Galleria 2, 6928 Manno, Switzerland. http://www.vetta.org/documents/IDSIA-12-06-1.pdf 3. M. Mahoney (2000), A Note on Lexical Acquisition in Text without Spaces, http://cs.fit.edu/~mmahoney/dissertation/lex1.html -- Matt Mahoney, [EMAIL PROTECTED] --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by