Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
2008/9/5 Mike Tintner [EMAIL PROTECTED]: 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. Not necessarily, the task could be interrupted at that process stopped or paused indefinately. 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. If it hung because of mult-process issues, you would need perfect knowledge of the environment to know the possible timing issues as well. 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. Computers aren't free thinkers, but it does not follow from an inability to switch, cancel, pause and restart or modify tasks. All of which they can do admirably. They just don't tend to do so, because they aren't smart enough (and cannot change themselves to be so) to know when it might be appropriate for what they are trying to do, so it is left up to the human operator to do so. I'm very interested in computers that self-maintain, that is reduce (or eliminate) the need for a human to be in the loop or know much about the internal workings of the computer. However it doesn't need a vastly different computing paradigm it just needs a different way of thinking about the systems. E.g. how can you design a system that does not need a human around to fix mistakes, upgrade it or maintain it in general. As they change their own system I will not know what they are going to do, because they can get information from the environment about how to act. This will me it a 'free thinker' of sorts. Whether it will be enough to get what you want, is an empirical matter, as far as I am concerned. Will --- 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
Will, Yes, humans are manifestly a RADICALLY different machine paradigm- if you care to stand back and look at the big picture. Employ a machine of any kind and in general, you know what you're getting - some glitches (esp. with complex programs) etc sure - but basically, in general, it will do its job. Humans are only human, not a machine. Employ one of those, incl. yourself, and, by comparison, you have only a v. limited idea of what you're getting - whether they'll do the job at all, to what extent, how well. Employ a programmer, a plumber etc etc.. Can you get a good one these days?... VAST difference. And that's the negative side of our positive side - the fact that we're 1) supremely adaptable, and 2) can tackle those problems that no machine or current AGI - (actually of course, there is no such thing at the mo, only pretenders) - can even *begin* to tackle. Our unreliability . That, I suggest, only comes from having no set structure - no computer program - no program of action in the first place. (Hey, good idea, who needs a program?) Here's a simple, extreme example. Will, I want you to take up to an hour, and come up with a dance, called the Keyboard Shuffle. (A very ill-structured problem.) Hey, you can do that. You can tackle a seriously ill-structured problem. You can embark on an activity you've never done before, presumably had no training for, have no structure for, yet you will, if cooperative, come up with something - cobble together a session of that activity, and end-product, an actual dance. May be shit, but it'll be a dance. And that's only an extreme example of how you approach EVERY activity. You similarly don't have a structure for your next hour[s], if you're writing an essay, or a program, or spending time watching TV, flipping chanels. You may quickly *adopt* or *form* certain structures/ routines. But they only go part way, and you do have to adopt and/or create them. Now, I assert, that's what an AGI is - a machine that has no programs, (no preset, complete structures for any activities), designed to tackle ill-structured problems by creating and adopting structures, not automatically following ones that have been laboured over for ridiculous amounts of time by human programmers offstage. And that in parallel, though in an obviously more constrained way, is what every living organism is - an extraordinary machine that builds itself adaptively and flexibly, as it goes along - Dawkins' famous plane that builds itself in mid-air. Just as we construct our activities in mid-air. Also a very different machine paradigm to any we have at the mo (although obviously lots of people are currently trying to design/understand such self-building machines). P.S. The irony is that scientists and rational philosophers, faced with the extreme nature of human imperfection - our extreme fallibility (in the sense described above - i.e. liable to fail/give up/procrastinate at any given activity at any point in a myriad of ways) - have dismissed it as, essentially, down to bugs in the system. Things that can be fixed. AGI-ers have the capacity like no one else to see and truly appreciate that such fallibility = highly desirable adaptability and that humans/animals really are fundamentally different machines. P.P.S. BTW that's the proper analogy for constructing an AGI - not inventing the plane (easy-peasy), but inventing the plane that builds itself in mid-air, (whole new paradigm of machine- and mind- invention). Will: 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. Not necessarily, the task could be interrupted at that process stopped or paused indefinately. 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. If it hung because of mult-process issues, you would need perfect knowledge of the environment to know the possible timing issues as well. 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. Computers aren't free thinkers, but it does not follow from an inability to switch, cancel, pause and restart or modify tasks. All of which they can do admirably. They just don't tend to do so, because they aren't smart enough (and cannot change
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
Sorry - para Our unreliability .. should have contined.. Our unreliabilty is the negative flip-side of our positive ability to stop an activity at any point, incl. the beginning and completely change tack/ course or whole approach, incl. the task itself, and even completely contradict ourself. --- 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)
Thinking out loud here as I find the relationship between compression and intelligence interesting: Compression in itself has the overriding goal of reducing storage bits. Intelligence has coincidental compression. There is resource management there. But I do think that it is not ONLY coincidental. Knowledge has structure which can be organized and naturally can collapse into a lower complexity storage state. Things have order, based on physics and other mathematical relationships. The relationship between compression and stored knowledge and intelligence is intriguing. But knowledge can be compressed inefficiently to where it inhibits extraction and other operations so there are differences with compression and intelligence related to computational expense. Optimal intelligence would have a variational compression structure IOW some stuff needs fast access time with minimal decompression resource expenditure and other stuff has high storage priority but computational expense and access time are not a priority. And then when you say the word compression there is a complicity of utility. The result of a compressor that has general intelligence still has a goal of reducing storage bits. I think that compression can be a byproduct of the stored knowledge created by a general intelligence. But if you have a compressor with general intelligence built in and you assign it a goal of taking input data and reducing the storage space it still may result in a series of hacks because that may be the best way of accomplishing that goal. Sure there may be some new undiscovered hacks that require general intelligence to uncover. And a compressor that is generally intelligent may produce more rich lossily compressed data from varied sources. The best lossy compressor is probably generally intelligent. They are very similar as you indicate... but when you start getting real lossy, when you start asking questions from your lossy compressed data that are not related to just the uncompressed input there is a difference there. Compression itself is just one dimensional. Intelligence is multi. John -Original Message- From: Matt Mahoney [mailto:[EMAIL PROTECTED] Sent: Friday, September 05, 2008 6:39 PM To: agi@v2.listbox.com Subject: 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
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
2008/9/6 Mike Tintner [EMAIL PROTECTED]: Will, Yes, humans are manifestly a RADICALLY different machine paradigm- if you care to stand back and look at the big picture. Employ a machine of any kind and in general, you know what you're getting - some glitches (esp. with complex programs) etc sure - but basically, in general, it will do its job. What exactly is a desktop computers job? Humans are only human, not a machine. Employ one of those, incl. yourself, and, by comparison, you have only a v. limited idea of what you're getting - whether they'll do the job at all, to what extent, how well. Employ a programmer, a plumber etc etc.. Can you get a good one these days?... VAST difference. If you find a new computer that I do not know how it has been programmed (whether it has linux/windows and what version). You also lack knowledge of what it is going to do. Aibo is a computer as well! It follows a program. And that's the negative side of our positive side - the fact that we're 1) supremely adaptable, and 2) can tackle those problems that no machine or current AGI - (actually of course, there is no such thing at the mo, only pretenders) - can even *begin* to tackle. Our unreliability . That, I suggest, only comes from having no set structure - no computer program - no program of action in the first place. (Hey, good idea, who needs a program?) You equate set structure with computer program. A computer program is not set! There is set structure of some sorts in the brain, at the neural level anyway. so you would have to be more precise in what you mean by lack of set structure. Wait, program of action? You don't think computer programs are like lists of things to do in the real world, do you? That is just something cooked up by the language writers to make things easier to deal with, a computer program is really only about memory manipulation. Some of the memory locations might be hooked up to the real world, but at the end of the day the computer treats it all as semanticless memory manipulations. Since what controls the memory manipulations are themselves in memory, they to can be manipulated! Here's a simple, extreme example. Will, I want you to take up to an hour, and come up with a dance, called the Keyboard Shuffle. (A very ill-structured problem.) How about you go learn about self-modifying assembly language, preferably with real-time interrupts. That would be a better use of the time, I think. 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
It has been explained many times to Tintner that even though computer hardware works with a particular set of primitive operations running in sequence, a hardwired set of primitive logical operations operating in sequence is NOT the theory of intelligence that any AGI researchers are proposing (to my knowledge). A computer is just a system for holding a theory of intelligence which does not look like those primitives (at least not since the view that intelligence consists of simple interpretations of atomic tokens representing physical objects in small numbers of relationships with other such tokens was given up decades ago as insufficient). As an example, the representational mechansms in Novamente and the dynamics of the mind agents that operate on them are probably better thought of as churning masses of probability relationships with varying and often non-specific semantic interpretations than Tintner's narrow view of what a computer is -- although I do not yet understand Novamente in detail. He has to ignore all such efforts, though, because if he paid attention he would have to stop saying that NONE of us understand ANYTHING about how REAL intelligence is actually based on line drawings, or keyboards, or other childish notions. Though he's in my killfile I do see his posts when others take the bait. So Mike, please try to finally understand this: AGI researchers do not think of intelligence as what you think of as a computer program -- some rigid sequence of logical operations programmed by a designer to mimic intelligent behavior. We know it is deeper than that. This has been clear to just about everybody for many many years. By engaging the field at such a level you do nothing worthwhile. Date: Sat, 6 Sep 2008 15:38:59 +0100 From: [EMAIL PROTECTED] To: agi@v2.listbox.com Subject: Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser 2008/9/6 Mike Tintner [EMAIL PROTECTED]: Will, Yes, humans are manifestly a RADICALLY different machine paradigm- if you care to stand back and look at the big picture. Employ a machine of any kind and in general, you know what you're getting - some glitches (esp. with complex programs) etc sure - but basically, in general, it will do its job. What exactly is a desktop computers job? --- 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: AI isn't cheap (was Re: Real vs. simulated environments (was Re: [agi] draft for comment.. P.S.))
Matt, I heartily disagree with your view as expressed here, and as stated to my by heads of CS departments and other high ranking CS PhDs, nearly (but not quite) all of whom have lost the fire in the belly that we all once had for CS/AGI. I DO agree that CS is like every other technological endeavor, in that almost everything that can be done as a PhD thesis has already been done. but there is a HUGE gap between a PhD thesis scale project and what that same person can do with another few more millions and a couple more years, especially if allowed to ignore the naysayers. The reply is a even more complex than your well documented statement, but I'll take my best shot at it, time permitting. Here, the angel is in the details. On 9/5/08, Matt Mahoney [EMAIL PROTECTED] wrote: --- 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], ... but we STILL do not have any sort of useful *unsupervised* NN, the equivalent of which seems to be needed for any good AGI. Note my recent postings about a potential theory of everything that would most directly hit unsupervised NN, providing not only a good way of operating these, but possibly the provably best way of operating. natural language processing and language translation [2], My Dr. Eliza is right there and showing that useful understanding out of precise context is almost certainly impossible. I regularly meet with the folks working on the Russian translator project, and rest assured, things are STILL advancing fairly rapidly. Here, there is continuing funding, and I expect that the Russian translator will eventually succeed (they already claim success). models of human decision making [3], These are curious but I believe them to be an emergent properties of processes that we don't understand at all, so they have no value other than for testing of future systems. Note that human decision making does NOT generally include many advanced sorts of logic that simply don't occur to ordinary humans, which is where an AGI could shine. Hence, understanding the human but not the non-human processes is nearly worthless. automatic theorem proving [4,8,10], Great for when you already have the answer - but what is it good for?! natural language databases [5], Which are only useful if/when the provably false presumption is true that NL understanding is generally possible. game playing programs [9,13], Note relevant for AGI. optical character recognition [14], Only recently have methods emerged that are truly font-independent. This SHOULD have been accomplished long ago (like shortly after your 1960 reference), but no one wanted to throw significant money at it. I nearly launched an OCR company (Cognitext) in 1981, but funding eventually failed * because* I had done the research and had a new (but *un*proven) method that was truly font-independent. handwriting and speech recognition [15], ... both of which are now good enough for AI interaction (e.g. my Gracie speech I/O interface to Dr. Eliza), but NOT good enough for general dictation. Unfortunately, the methods used don't seem to shed much light on how the underlying processes work in us. and important theoretical work [16,17,18]. Note again my call for work/help on what I call computing's theory of everything leveraging off of principal component analysis. Since then we have had mostly just incremental improvements. YES. This only shows that the support process has long been broken. and NOT that there isn't a LOT of value that is just out of reach of PhD-sized projects. Big companies like Google and Microsoft have strong incentives to develop AI Internal politics at both (that I have personally run into) restrict expenditures to PROVEN methods, as a single technical failure spells doom for the careers of everyone working on them. Hence, their RD is all D and no R. and have billions to spend. Not one dollar of which goes into what I would call genuine research. Maybe the problem really is hard. ... and maybe it is just a little difficult. My own Dr. Eliza program has seemingly unbelievable NL-stated problem solving capabilities, but is built mostly on the same sort of 1960s technology you cited. Why wasn't it built before 1970? I see two simple reasons: 1. Joe Weizenbaum, in his *Computer Power and Human Reason,* explained why this approach could never work. That immediately made it impossible to get any related effort funded or acceptable in a university setting. 2. It took about a year to make a demonstrable real-world NL problem solving system, which would have been at the outer reaches of a PhD or casual personal project. I have
Re: Language modeling (was Re: [agi] draft for comment)
--- On Fri, 9/5/08, Pei Wang [EMAIL PROTECTED] wrote: 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. I don't advocate the Turing test as the ideal test of intelligence. Turing himself was aware of the problem when he gave an example of a computer answering an arithmetic problem incorrectly in his famous 1950 paper: Q: Please write me a sonnet on the subject of the Forth Bridge. A: Count me out on this one. I never could write poetry. Q: Add 34957 to 70764. A: (Pause about 30 seconds and then give as answer) 105621. Q: Do you play chess? A: Yes. Q: I have K at my K1, and no other pieces. You have only K at K6 and R at R1. It is your move. What do you play? A: (After a pause of 15 seconds) R-R8 mate. I prefer a preference test, which a machine passes if you prefer to talk to it over a human. Such a machine would be too fast and make too few errors to pass a Turing test. For example, if you had to add two large numbers, I think you would prefer to use a calculator than ask someone. You could, I suppose, measure intelligence as the fraction of questions for which the machine gives the preferred answer, which would be 1/4 in Turing's example. If you know the probability distribution P of text, and therefore know the distribution P(A|Q) for any question Q and answer A, then to pass the Turing test you would randomly choose answers from this distribution. But to pass the preference test for all Q, you would choose A that maximizes P(A|Q) because the most probable answer is usually the correct one. Text compression measures progress toward either test. I believe that compression measures your definition of intelligence, i.e. adaptation given insufficient knowledge and resources. In my benchmark, there are two parts: the size of the decompression program, which measures the initial knowledge, and the compressed size, which measures prediction errors that occur as the system adapts. Programs must also meet practical time and memory constraints to be listed in most benchmarks. Compression is also consistent with Legg and Hutter's universal intelligence, i.e. expected reward of an AIXI universal agent in an environment simulated by a random program. Suppose you have a compression oracle that inputs any string x and outputs the shortest program that outputs a string with prefix x. Then this reduces the (uncomputable) AIXI problem to using the oracle to guess which environment is consistent with the interaction so far, and figuring out which future outputs by the agent will maximize reward. Of course universal intelligence is also not testable because it requires an infinite number of environments. Instead, we have to choose a practical data set. I use Wikipedia text, which has fewer errors than average text, but I believe that is consistent with my goal of passing the preference test. -- 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 Sat, 9/6/08, John G. Rose [EMAIL PROTECTED] wrote: Compression in itself has the overriding goal of reducing storage bits. Not the way I use it. The goal is to predict what the environment will do next. Lossless compression is a way of measuring how well we are doing. -- 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)
I won't argue against your preference test here, since this is a big topic, and I've already made my position clear in the papers I mentioned. As for compression, yes every intelligent system needs to 'compress' its experience in the sense of keeping the essence but using less space. However, it is clearly not loseless. It is even not what we usually call loosy compression, because what to keep and in what form is highly context-sensitive. Consequently, this process is not reversible --- no decompression, though the result can be applied in various ways. Therefore I prefer not to call it compression to avoid confusing this process with the technical sense of compression, which is reversible, at least approximately. Legg and Hutter's universal intelligence definition is way too narrow to cover various attempts towards AI, even as an idealization. Therefore, I don't take it as a goal to aim at and to approach to as close as possible. However, as I said before, I'd rather leave this topic for the future, when I have enough time to give it a fair treatment. Pei On Sat, Sep 6, 2008 at 4:29 PM, Matt Mahoney [EMAIL PROTECTED] wrote: --- On Fri, 9/5/08, Pei Wang [EMAIL PROTECTED] wrote: 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. I don't advocate the Turing test as the ideal test of intelligence. Turing himself was aware of the problem when he gave an example of a computer answering an arithmetic problem incorrectly in his famous 1950 paper: Q: Please write me a sonnet on the subject of the Forth Bridge. A: Count me out on this one. I never could write poetry. Q: Add 34957 to 70764. A: (Pause about 30 seconds and then give as answer) 105621. Q: Do you play chess? A: Yes. Q: I have K at my K1, and no other pieces. You have only K at K6 and R at R1. It is your move. What do you play? A: (After a pause of 15 seconds) R-R8 mate. I prefer a preference test, which a machine passes if you prefer to talk to it over a human. Such a machine would be too fast and make too few errors to pass a Turing test. For example, if you had to add two large numbers, I think you would prefer to use a calculator than ask someone. You could, I suppose, measure intelligence as the fraction of questions for which the machine gives the preferred answer, which would be 1/4 in Turing's example. If you know the probability distribution P of text, and therefore know the distribution P(A|Q) for any question Q and answer A, then to pass the Turing test you would randomly choose answers from this distribution. But to pass the preference test for all Q, you would choose A that maximizes P(A|Q) because the most probable answer is usually the correct one. Text compression measures progress toward either test. I believe that compression measures your definition of intelligence, i.e. adaptation given insufficient knowledge and resources. In my benchmark, there are two parts: the size of the decompression program, which measures the initial knowledge, and the compressed size, which measures prediction errors that occur as the system adapts. Programs must also meet practical time and memory constraints to be listed in most benchmarks. Compression is also consistent with Legg and Hutter's universal intelligence, i.e. expected reward of an AIXI universal agent in an environment simulated by a random program. Suppose you have a compression oracle that inputs any string x and outputs the shortest program that outputs a string with prefix x. Then this reduces the (uncomputable) AIXI problem to using the oracle to guess which environment is consistent with the interaction so far, and figuring out which future outputs by the agent will maximize reward. Of course universal intelligence is also not testable because it requires an infinite number of environments. Instead, we have to choose a practical data set. I use Wikipedia text, which has fewer errors than average text, but I believe that is consistent with my goal of passing the preference test. -- 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 Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed:
Re: AI isn't cheap (was Re: Real vs. simulated environments (was Re: [agi] draft for comment.. P.S.))
Steve, where are you getting your cost estimate for AGI? Is it a gut feeling, or something like the common management practice of I can afford $X so it will cost $X? My estimate of $10^15 is based on the value of the world economy, US $66 trillion per year and increasing 5% annually over the next 30 years, which is how long it will take for the internet to grow to the computational power of 10^10 human brains (at 10^15 bits and 10^16 OPS each) at the current rate of growth, doubling every couple of years. Even if you disagree with these numbers by a factor of 1000, it only moves the time to AGI by a few years, so the cost estimate hardly changes. And even if the hardware is free, you still have to program or teach about 10^16 to 10^17 bits of knowledge, assuming 10^9 bits of knowledge per brain [1] and 1% to 10% of this is not known by anyone else. Software and training costs are not affected by Moore's law. Even if we assume human level language understanding and perfect sharing of knowledge, the training cost will be 1% to 10% of your working life to train the AGI to do your job. Also, we have made *some* progress toward AGI since 1965, but it is mainly a better understanding of why it is so hard, e.g. - We know that general intelligence is not computable [2] or provable [3]. There is no neat theory. - From Cyc, we know that coding common sense is more than a 20 year effort. Lenat doesn't know how much more, but guesses it is maybe between 0.1% and 10% finished. - Google is the closest we have to AI after a half trillion dollar effort. 1. Landauer, Tom (1986), “How much do people remember? Some estimates of the quantity of learned information in long term memory”, Cognitive Science (10) pp. 477-493. 2. Hutter, Marcus (2003), A Gentle Introduction to The Universal Algorithmic Agent {AIXI}, in Artificial General Intelligence, B. Goertzel and C. Pennachin eds., Springer. http://www.idsia.ch/~marcus/ai/aixigentle.htm 3. 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 -- Matt Mahoney, [EMAIL PROTECTED] --- On Sat, 9/6/08, Steve Richfield [EMAIL PROTECTED] wrote: From: Steve Richfield [EMAIL PROTECTED] Subject: Re: AI isn't cheap (was Re: Real vs. simulated environments (was Re: [agi] draft for comment.. P.S.)) To: agi@v2.listbox.com Date: Saturday, September 6, 2008, 2:58 PM Matt, I heartily disagree with your view as expressed here, and as stated to my by heads of CS departments and other high ranking CS PhDs, nearly (but not quite) all of whom have lost the fire in the belly that we all once had for CS/AGI. I DO agree that CS is like every other technological endeavor, in that almost everything that can be done as a PhD thesis has already been done. but there is a HUGE gap between a PhD thesis scale project and what that same person can do with another few more millions and a couple more years, especially if allowed to ignore the naysayers. The reply is a even more complex than your well documented statement, but I'll take my best shot at it, time permitting. Here, the angel is in the details. On 9/5/08, Matt Mahoney [EMAIL PROTECTED] wrote: --- 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], ... but we STILL do not have any sort of useful unsupervised NN, the equivalent of which seems to be needed for any good AGI. Note my recent postings about a potential theory of everything that would most directly hit unsupervised NN, providing not only a good way of operating these, but possibly the provably best way of operating. natural language processing and language translation [2], My Dr. Eliza is right there and showing that useful understanding out of precise context is almost certainly impossible. I regularly meet with the folks working on the Russian translator project, and rest assured, things are STILL advancing fairly rapidly. Here, there is continuing funding, and I expect that the Russian translator will eventually succeed (they already claim success). models of human decision making [3], These are curious but I believe them to be an emergent properties of processes that we don't understand at all, so they have no value other than for testing of future systems. Note that human decision making does NOT generally include many advanced sorts of logic that simply don't occur to ordinary humans, which is where an AGI could
Re: [agi] A NewMetaphor for Intelligence - the Computer/Organiser
DZ:AGI researchers do not think of intelligence as what you think of as a computer program -- some rigid sequence of logical operations programmed by a designer to mimic intelligent behavior. 1. Sequence/Structure. The concept I've been using is not that a program is a sequence of operations but a structure., including as per NARS, as I 've read Pei, a structure that may change more or less continuously. Techno-idiot that I am, I am fairly aware that many modern programs are extremely sophisticated and complex structures. I take into account, for example, Minsky's idea of a possible society of mind, with many different parts perhaps competing - not obviously realised in program form yet. But programs are nevertheless manifestly structures. Would you dispute that? And a central point I've been making is that human life and activities are manifestly *unstructured* - that in just about everything we do, we struggle to impose structure on our activities - to impose order and organization., planning, focus etc. . Especially in AGI's central challenge -creativity. Creative activities are outstanding examples of unstructured activities, in which structures have to be created - painting scenes, writing stories, designing new machines, writing music/pop songs - often starting from an entirely blank page. (What's the program equivalent?) 2. A Programmer on Programs. I am persuaded on multiple grounds that the human mind is not always algorithmic, nor merely computational in the syntactic sense of computational. S Kauffman, Reinventing the Sacred Try Chap 12. Computationally, he trumps most AGI-ers in terms of most AI departments, incl. complexity, bioinformatics and general standing, no? Read the whole book in fact - it can be read as being entirely about the creative problem/challenge of AGI - you liked Barsalou, you'll like this. . --- 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 Sat, 9/6/08, Pei Wang [EMAIL PROTECTED] wrote: As for compression, yes every intelligent system needs to 'compress' its experience in the sense of keeping the essence but using less space. However, it is clearly not loseless. It is even not what we usually call loosy compression, because what to keep and in what form is highly context-sensitive. Consequently, this process is not reversible --- no decompression, though the result can be applied in various ways. Therefore I prefer not to call it compression to avoid confusing this process with the technical sense of compression, which is reversible, at least approximately. I think you misunderstand my use of compression. The goal is modeling or prediction. Given a string, predict the next symbol. I use compression to estimate how accurate the model is. It is easy to show that if your model is accurate, then when you connect your model to an ideal coder (such as an arithmetic coder), then compression will be optimal. You could actually skip the coding step, but it is cheap, so I use it so that there is no question of making a mistake in the measurement. If a bug in the coder produces a too small output, then the decompression step won't reproduce the original file. In fact, many speech recognition experiments do skip the coding step in their tests and merely calculate what the compressed size would be. (More precisely, they calculate word perplexity, which is equivalent). The goal of speech recognition is to find the text y that maximizes P(y|x) for utterance x. It is common to factor the model using Bayes law: P(y|x) = P(x|y)P(y)/P(x). We can drop P(x) since it is constant, leaving the acoustic model P(x|y) and language model P(y) to evaluate. We know from experiments that compression tests on P(y) correlate well with word error rates for the overall system. Internally, all lossless compressors use lossy compression or data reduction to make predictions. Most commonly, a context is truncated and possibly hashed before looking up the statistics for the next symbol. The top lossless compressors in my benchmark use more sophisticated forms of data reduction, such as mapping upper and lower case letters together, or mapping groups of semantically or syntactically related words to the same context. As a test, lossless compression is only appropriate for text. For other hard AI problems such as vision, art, and music, incompressible noise would overwhelm the human-perceptible signal. Theoretically you could compress video to 2 bits per second (the rate of human long term memory) by encoding it as a script. The decompressor would read the script and create a new movie. The proper test would be lossy compression, but this requires human judgment to evaluate how well the reconstructed data matches the original. -- 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