Re: [agi] What best evidence for fast AI?
On Saturday 10 November 2007 16:51, Robin Hanson wrote: At 02:06 PM 11/10/2007, Richard Loosemore wrote: Basically, 'traditional' AI people have an almost theological aversion to the idea that the task of building an AI might involve having to learn (and deconstruct!) a vast amount of cognitive science, and then use an experimental-science methodology to find the mechanisms that really give rise to AI. I have to give a lot of weight to the apparent fact that most AI researchers have not yet been convinced to accept your favored approach. More persuasive to me are arguments for fast AI based on more widely shared premises. I believe that both Richard and Robin misrepresent the profession when they reference traditional AI researchers. I believe that they are thinking only of those researchers who have concluded that AI is possible with today's technology without further advances in cognitive science being required. Thus, only those who believe in such a thing are included in the group. I think this unfairly excludes the vast larger number of computer experts, cognitive experts, and those with knowledge in both fields, who have studied the field of AI and have concluded that such a thing is not possible at this time. Admitidly, most computer scientists and cognitive experts do not agree with the approach being discussed above. Therefore, using Robin's reasoning, I would have to give more weight to all these people, instead of assuming that they are all wrong, and that the small minority of our favorite AI researchers are correct. Therefore, I use Robin's logic to agree with Richard's conclusion! -- Harvey Newstrom CISSP CISA CISM CIFI NSA-IAM GSEC ISSAP ISSMP ISSPCS IBMCP - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=66361805-203259
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Mike Tintner wrote: Sounds a little confusing. Sounds like you plan to evolve a system through testing thousands of candidate mechanisms. So one way or another you too are taking a view - even if it's an evolutionary, I'm not taking a view view - on, and making a lot of asssumptions about -how systems evolve -the known architecture of human cognition. No, I think because of the paucity of information I gave you have misunderstood slightly. Everything I mentioned was in the context of an extremely detailed framework that tries to include all of the knowledge we have so far gleaned by studying human cognition using the methods of cognitive science. So I am not making assumptions about the architecture of human cognition I am using every scrap of experimental data I can. You can say that this is still assuming that the framework is correct, but that is nothing compared to the usual assumptions made in AI, where the programmer just picks up a grab bag of assorted ideas that are floating around in the literature (none of them part of a coherent theory of cognition) and starts hacking. And just because I talk of thousands of candidate mechanisms, that does not mean that there is evolution involved: it just means that even with a complete framework for human cognition to start from there are still so many questions about the low-level to high-level linkage that a vast number of mechanisms have to be explored. about which science has extremely patchy and confused knowledge. I don't see how any system-builder can avoid taking a view of some kind on such matters, yet you seem to be criticising Ben for so doing. Ben does not start from a complete framework for human cognition, nor does he feel compelled to stick close to the human model, and my criticisms (at least in this instance) are not really about whether or not he has such a framework, but about a problem that I can see on his horizon. I was hoping that you also had some view on how a system 's symbols should be grounded, especially since you mention Harnad, who does make vague gestures towards the brain's levels of grounding. But you don't indicate any such view. On the contrary, I explained exactly how they would be grounded: if the system is allowed to build its own symbols *without* me also inserting ungrounded (i.e. interpreted, programmer-constructed) symbols and messing the system up by forcing it to use both sorts of symbols, then ipso fact it is grounded. It is easy to build a grounded system. The trick is to make it both grounded and intelligent at the same time. I have one strategy for ensuring that it turns out intelligent, and Ben has another my problem with Ben's strategy is that I believe his attempt to ensure that the system is intelligent ends up compromising the groundedness of the system. Sounds like you too, pace MW, are hoping for a number of miracles - IOW creative ideas - to emerge, and make your system work. I don't understand where I implied this. You have to remember that I am doing this within a particular strategy (outlined in my CSP paper). When you see me exploring 'thousands' of candidate mechanisms to see how one parameter plays a role, this is not waiting for a miracle, it is a vital part of the strategy. A strategy that, I claim, is the only viable one. Anyway, you have to give Ben credit for putting a lot of his stuff principles out there on the line. I think anyone who wants to mount a full-scale assault on him ( why not?) should be prepared to reciprocate. Nice try, but there are limits to what I can do to expose the details. I have not yet worked out how much I should release and how much to withhold (I confess, I nearly decided to go completely public a month or so ago, but then changed my mind after seeing the dismally poor response that even one of the ideas provoked). Maybe in the near future I will write a summary account. In the mean time, yes, it is a little unfair of me to criticise other projects. But not that unfair. When a scientist sees a big problem with a theory, do you suppose they wait until they have a completely worked out alternative before discussing the fact that there is a problem with the theory that other people may be praising? That is not the way of science. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65349870-56ef76
Re: [agi] What best evidence for fast AI?
The problem with probability-based conflict resolution is that it is a hack to get around insufficient knowledge rather than an attempt to figure out how to get more knowledge ED This agrees with what I said above about not putting enough emphasis on selecting what probabilistic formulas are appropriate. But it doesn't argue against the importance of probabilities It argues against using them blindly. ED So by operating with small amounts of data how small, very roughly, are you talking about. And are you only talking about the active goals or sources of activation, that will be small or are you saying that all the computation in the system will only be dealing with a small amount of data within, for example, one second of the processing of human-level system operating at human-level speed? MARK I mean like the way humans reason, there is only concentration on a small number of objects -- which are only one link away from an almost inconceivable number of related things -- and then the brain can jump at least three of these links with lightning rapidity. ED So this implies you are not arguing against the idea that AGI will be dealing with massive data, just that that use will be focused by a concentration on a relatively small number of sources of activation at once. MARK Ask Ben how much actual work has been done on activation control in very large, very sparse atom spaces in Novamente. He'll tell you that it's a project for when he's further along. I'll insist (as will Richard) that if it isn't baked in from the very beginning, you're probably going to have to go back to the beginning to repair the lack. ED It is exactly such research I want to see funded. It strikes me as one of the key things we must learn to do well to make powerful AGI. But I think even with some fairly dumb activation control systems you could get useful results. Such results would not be at all human-level in may ways, but in other ways they could be much more powerful because such systems could deal with many more explicit facts and could input and output information at a much higher rate than humans. For example, what is the equivalent of the activation control (or search) algorithm in Google sets. They operate over huge data. I bet the algorithm for calculating their search or activation is relatively simple (much, much, much less than a PhD theses) and look what they can do. So I think one path is to come up with applications that can use and reason with large data, having roughly world knowledge-like sparseness, (such as NL data) and start with relatively simple activation algorithms and develop then from the ground up. MARK P.S. Oh yeah -- if you were public enemy number one, I wouldn't bother answering you (and I probably should lay off of the fan-boy crap :-). ED Thanks. I admit I am impressed with Novamente. Since it's the best AGI architecture I currently know of; I am impressed with Ben; believe there is a high probability all the gaps you address could be largely fixed within five years with deep funding (which may never come); and since I want to get such deep funding for just the type of large atom-base work you say is so critical, I think it is important to focus on the potential for greatness that Novamente and somewhat similar systems have, rather than only think of its current gaps and potential problems. But of course, at the same time, we must look for and try to understand its gaps and potential problems so that we can remove them. Ed Porter -Original Message- From: Mark Waser [mailto:[EMAIL PROTECTED] Sent: Monday, November 12, 2007 2:42 PM To: agi@v2.listbox.com Subject: Re: [agi] What best evidence for fast AI? It is NOT clear that Novamente documentation is NOT enabling, or could not be made enabling, with, say, one man year of work. Strong argument could be made both ways. I believe that Ben would argue that Novamente documentation is NOT enabling even with one man-year of work. Ben? There is still way to much *research* work to be done. But the standard for non-enablement is very arguably weaker than not requiring a miracle. It would be more like not requiring a leap of creativity that is outside the normal skill of talented PhDs trained in related fields. So although your position is reasonable, I hope you understand so is that on the other side. My meant-to-be-humorous miracle phrasing is clearly throwing you. The phrase not requiring a leap of creativity that is outside the normal skill of talented PhDs trained in related fields works for me. Novamente is *definitely* not there yet. I'm rather sure that Ben would agree -- as in, I'm not on the other side, *you* are on the other side from the system's designer. Again, Ben please feel free to chime in. much scaling stuff Remember that the brain is *massively* parallel
RE: [agi] What best evidence for fast AI?
Lukasz, Which of the multiple issues that Mark listed is one of the two basic directions you were referring to. Ed Porter -Original Message- From: Lukasz Stafiniak [mailto:[EMAIL PROTECTED] Sent: Wednesday, November 14, 2007 9:15 AM To: agi@v2.listbox.com Subject: Re: [agi] What best evidence for fast AI? I think that there are two basic directions to better the Novamente architecture: the one Mark talks about more integration of MOSES with PLN and RL theory On 11/13/07, Edward W. Porter [EMAIL PROTECTED] wrote: Response to Mark Waser Mon 11/12/2007 2:42 PM post. MARK Remember that the brain is *massively* parallel. Novamente MARK and any other linear (or minorly-parallel) system is *not* going to work in the same fashion as the brain. Novamente can be parallelized to some degree but *not* to anywhere near the same degree as the brain. I love your speculation and agree with it -- but it doesn't match near-term reality. We aren't going to have brain-equivalent parallelism anytime in the near future. ED I think in five to ten years there could be computers capable ED of providing every bit as much parallelism as the brain at prices that will allow thousands or hundreds of thousands of them to be sold. But it is not going to happen overnight. Until then the lack of brain level hardware is going to limit AGI. But there are still a lot of high value system that could be built on say $100K to $10M of hardware. You claim we really need experience with computing and controlling activation over large atom tables. I would argue that obtaining such experience should be a top priority for government funders. MARK The node/link architecture is very generic and can be used MARK for virtually anything. There is no rational way to attack it. It is, I believe, going to be the foundation for any system since any system can easily be translated into it. Attacking the node/link architecture is like attacking assembly language or machine code. Now -- are you going to write your AGI in assembly language? If you're still at the level of arguing node/link, we're not communicating well. ED nodes and links are what patterns are made of, and each static pattern can have an identifying node associated with it as well as the nodes and links representing its sub-patterns, elements, the compositions of which it is part, it associations, etc. The system automatically organize patterns into a gen/comp hierarchy. So, I am not just dealing at a node and link level, but they are the basic building blocks. MARK ... I *AM* saying that the necessity of using probabilistic reasoning for day-to-day decision-making is vastly over-rated and has been a horrendous side-road for many/most projects because they are attempting to do it in situations where it is NOT appropriate. The increased, almost ubiquitous adaptation of probabilistic methods is the herd mentality in action (not to mention the fact that it is directly orthogonal to work thirty years older). Most of the time, most projects are using probabilistic methods to calculate a tenth place decimal of a truth value when their data isn't even sufficient for one. If you've got a heavy-duty discovery system, probabilistic methods are ideal. If you're trying to derive probabilities from a small number of English statements (like this raven is white and most ravens are black), you're seriously on the wrong track. If you go on and on about how humans don't understand Bayesian reasoning, you're both correct and clueless in not recognizing that your very statement points out how little Bayesian reasoning has to do with most general intelligence. Note, however, that I *do* believe that probabilistic methods *are* going to be critically important for activation for attention, etc. ED I agree that many approaches accord too much importance to the numerical accuracy and Bayesian purity of their approach, and not enough importance on the justification for the Bayesian formulations they use. I know of one case where I suggested using information that would almost certainly have improved a perception process and the suggestion was refused because it would not fit within the system's probabilistic framework. At an AAAI conference in 1997 I talked to a programmer for a big defense contractor who said he as a fan of fuzzy logic system; that they were so much more simple to get up an running because you didn't have to worry about probabilistic purity. He said his group that used fuzzy logic was getting things out the door that worked faster than the more probability limited competition. So obviously there is something to say for not letting probabilistic purity get in the way of more reasonable approaches. But I still think probabilities are darn important. Even your this raven is white and most ravens are black example involves notions of probability. We attribute
Re: [agi] What best evidence for fast AI?
On Nov 14, 2007 3:48 PM, Edward W. Porter [EMAIL PROTECTED] wrote: Lukasz, Which of the multiple issues that Mark listed is one of the two basic directions you were referring to. Ed Porter (First of all, I'm sorry for attaching my general remark as a reply: I was writing from a cell-phone which limited navigation.) I think, that it would be a more fleshed-out knowledge representation (but without limiting the representation-building flexibility of Novamente). -Original Message- From: Lukasz Stafiniak [mailto:[EMAIL PROTECTED] Sent: Wednesday, November 14, 2007 9:15 AM To: agi@v2.listbox.com Subject: Re: [agi] What best evidence for fast AI? I think that there are two basic directions to better the Novamente architecture: the one Mark talks about more integration of MOSES with PLN and RL theory - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64970556-f74c23
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Linas Vepstas wrote: On Tue, Nov 13, 2007 at 12:34:51PM -0500, Richard Loosemore wrote: Suppose that in some significant part of Novamente there is a representation system that uses probability or likelihood numbers to encode the strength of facts, as in [I like cats](p=0.75). The (p=0.75) is supposed to express the idea that the statement [I like cats] is in some sense 75% true. Either way, we have a problem: a fact like [I like cats](p=0.75) is ungrounded because we have to interpret it. Does it mean that I like cats 75% of the time? That I like 75% of all cats? 75% of each cat? Are the cats that I like always the same ones, or is the chance of an individual cat being liked by me something that changes? Does it mean that I like all cats, but only 75% as much as I like my human family, which I like(p=1.0)? And so on and so on. Eh? You are standing at the proverbial office water coooler, and Aneesh says Wen likes cats. On your drive home, you mind races .. does this mean that Wen is a cat fancier? You were planning on taking Wen out on a date, and this tidbit of information could be useful ... when you try to build the entire grounding mechanism(s) you are forced to become explicit about what these numbers mean, during the process of building a grounding system that you can trust to be doing its job: you cannot create a mechanism that you *know* is constructing sensible p numbers and facts during all of its development *unless* you finally bite the bullet and say what the p numbers really mean, in fully cashed out terms. But has a human, asking Wen out on a date, I don't really know what Wen likes cats ever really meant. It neither prevents me from talking to Wen, or from telling my best buddy that ...well, I know, for instance, that she likes cats... Lack of grounding is what makes humour funny, you can do a whole Pygmalion / Seinfeld episode on she likes cats. No: the real concept of lack of grounding is nothing so simple as the way you are using the word grounding. Lack of grounding makes an AGI fall flat on its face and not work. I can't summarize the grounding literature in one post. (Though, heck, I have actually tried to do that in the past: didn't do any good). Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64980585-67cbc9
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Hi, No: the real concept of lack of grounding is nothing so simple as the way you are using the word grounding. Lack of grounding makes an AGI fall flat on its face and not work. I can't summarize the grounding literature in one post. (Though, heck, I have actually tried to do that in the past: didn't do any good). FYI, I have read the symbol-grounding literature (or a lot of it), and generally found it disappointingly lacking in useful content... though I do agree with the basic point that non-linguistic grounding is extremely helpful for effective manipulation of linguistic entities... -- Ben G - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64981284-09925d
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Benjamin Goertzel wrote: On Nov 13, 2007 2:37 PM, Richard Loosemore [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote: Ben, Unfortunately what you say below is tangential to my point, which is what happens when you reach the stage where you cannot allow any more vagueness or subjective interpretation of the qualifiers, because you have to force the system to do its own grounding, and hence its own interpretation. I don't see why you talk about forcing the system to do its own grounding -- the probabilities in the system are grounded in the first place, as they are calculated based on experience. The system observes, records what it sees, abstracts from it, and chooses actions that it guess will fulfill its goals. Its goals are ultimately grounded in in-built feeling-evaluation routines, measuring stuff like amount of novelty observed, amount of food in system etc. So, the system sees and then acts ... and the concepts it forms and uses are created/used based on their utility in deriving appropriate actions. There is no symbol-grounding problem except in the minds of people who are trying to interpret what the system does, and get confused. Any symbol used within the system, and any probability calculated by the system, are directly grounded in the system's experience. There is nothing vague about an observation like Bob_Yifu was observed at time-stamp 599933322, or a fact Command 'wiggle ear' was sent at time-stamp 54. These perceptions and actions are the root of the probabilities the system calculated, and need no further grounding. What you gave below was a sketch of some more elaborate 'qualifier' mechanisms. But I described the process of generating more and more elaborate qualifier mechanisms in the body of the essay, and said why this process was of no help in resolving the issue. So, if a system can achieve its goals based on choosing procedures that it thinks are likely to achieve its goals, based on the knowledge it gathered via its perceived experience -- why do you think it has a problem? I don't really understand your point, I guess. I thought I did -- I thought your point was that precisely specifying the nature of a conditional probability is a rats-nest of complexity. And my response was basically that in Novamente we don't need to do that, because we define conditional probabilities based on the system's own knowledge-base, i.e. Inheritance A B .8 means If A and B were reasoned about a lot, then A would (as measred by an weighted average) have 80% of the relationships that B does But apparently you were making some other point, which I did not grok, sorry... Anyway, though, Novamente does NOT require logical relations of escalating precision and complexity to carry out reasoning, which is one thing you seemed to be assuming in your post. You are, in essence, using one of the trivial versions of what symbol grounding is all about. The complaint is not your symbols are not connected to experience. Everyone and their mother has an AI system that could be connected to real world input. The simple act of connecting to the real world is NOT the core problem. If you have an AGI system in which the system itself is allowed to do all the work of building AND interpreting all of its symbols, I don't have any issues with it. Where I do have an issue is with a system which is supposed to be doing the above experiential pickup, and where the symbols are ALSO supposed to be interpretable by human programmers who are looking at things like probability values attached to facts. When a programmer looks at a situation like ContextLink .7,.8 home InheritanceLink Bob_Yifu friend ... and then follows this with a comment like: which suggests that Bob is less friendly at home than in general. ... they have interpreted the meaning of that statement using their human knowledge. So here I am, looking at this situation, and I see: AGI system intepretation (implicit in system use of it) Human programmer intepretation and I ask myself which one of these is the real interpretation? It matters, because they do not necessarily match up. The human programmer's intepretation has a massive impact on the system because all the inference and other mechanisms are built around the assumption that the probabilities mean a certain set of things. You manipulate those p values, and your manipulations are based on assumptions about what they mean. But if the system is allowed to pick up its own knowledge from the environment, the implicit meaning of those p values will not necessarily match the human interpretation. As I say, the meaning is then implicit in the way the system *uses* those p values (and other stuff). It is a nontrivial question to ask whether the implicit system interpretation does indeed match the human intepretation built into the inference
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Benjamin Goertzel wrote: Hi, No: the real concept of lack of grounding is nothing so simple as the way you are using the word grounding. Lack of grounding makes an AGI fall flat on its face and not work. I can't summarize the grounding literature in one post. (Though, heck, I have actually tried to do that in the past: didn't do any good). FYI, I have read the symbol-grounding literature (or a lot of it), and generally found it disappointingly lacking in useful content... though I do agree with the basic point that non-linguistic grounding is extremely helpful for effective manipulation of linguistic entities... Ben, As you will recall, Harnad himself got frustrated with the many people who took the term symbol grounding and trivialized or distorted it in various ways. One of the reasons the grounding literature is such a waste of time (and you are right: it is) is that so many people talked so much nonsense about it. As far as I am concerned, your use of it is one of those trivial senses that Harnad complained of. (Essentially, if the system uses world input IN ANY WAY during the building of its symbols, then the system is grounded). The effort I put into that essay yesterday will have been completely wasted if your plan is to stick to that interpretation and not discuss the deeper issue that I raised. I really have no energy for pursuing yet another discussion about symbol grounding. Sorry: don't mean to blow you off, but you and I both have better things to do, and I foresee a big waste of time ahead if we pursue it. So let's just drop it? Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64998305-6bdb18
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Richard, So here I am, looking at this situation, and I see: AGI system intepretation (implicit in system use of it) Human programmer intepretation and I ask myself which one of these is the real interpretation? It matters, because they do not necessarily match up. That is true, but in some cases they may approximate each other well.. In others, not... This happens to be a pretty simple case, so the odds of a good approximation seem high. The human programmer's intepretation has a massive impact on the system because all the inference and other mechanisms are built around the assumption that the probabilities mean a certain set of things. You manipulate those p values, and your manipulations are based on assumptions about what they mean. Well, the PLN inference engine's treatment of ContextLink home InheritanceLink Bob_Yifu friend is in no way tied to whether the system's implicit interpretation of the ideas of home or friend are humanly natural, or humanly comprehensible. The same inference rules will be applied to cases like ContextLink Node_66655 InheritanceLink Bob_Yifu Node_544 where the concepts involved have no humanly-comprehensible label. It is true that the interpretation of ContextLink and InheritanceLink are fixed by the wiring of the system, in a general way (but what kinds of properties are referred to by them may vary in a way dynamically determined by the system). In order to completely ground the system, you need to let the system build its own symbols, yes, but that is only half the story: if you still have a large component of the system that follows a programmer-imposed interpretation of things like probability values attached to facts, you have TWO sets of symbol-using mechanisms going on, and the system is not properly grounded (it is using both grounded and ungrounded symbols within one mechanism). I don't think the system needs to learn its own probabilistic reasoning rules in order to be an AGI. This, to me, is too much like requiring that a brain needs to learn its own methods for modulating the conductances of the bundles of synapses linking between the neurons in cell assembly A and cell assembly B. I don't see a problem with the AGI system having hard-wired probabilistic inference rules, and hard-wired interpretations of probabilistic link types. But the interpretation of any **particular** probabilistic relationship inside the system, is relative to the concepts and the empirical and conceptual relationships that the system has learned. You may think that the brain learns its own uncertain inference rules based on a lower-level infrastructure that operates in terms entirely unconnected from ideas like uncertainty and inference. I think this is wrong. I think the brain's uncertain inference rules are the result, on the cell assembly level, of Hebbian learning and related effects on the neuron/synapse level. So I think the brain's basic uncertain inference rules are wired-in, just as Novamente's are, though of course using a radically different infrastructure. Ultimately an AGI system needs to learn its own reasoning rules and radically modify and improve itself, if it's going to become strongly superhuman! But that is not where we need to start... -- Ben - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64998317-8c4281
Re: [agi] What best evidence for fast AI?
Bryan Bishop wrote: On Tuesday 13 November 2007 09:11, Richard Loosemore wrote: This is the whole brain emulation approach, I guess (my previous comments were about evolution of brains rather than neural level duplication). Ah, you are right. But this too is an interesting topic. I think that the order of magnitudes for whole brain emulation, connectome, and similar evolutionary methods, are roughly the same, but I haven't done any calculations. It seems quite possible that what we need is a detailed map of every synapse, exact layout of dendritic tree structures, detailed knowledge of the dynamics of these things (they change rapidly) AND wiring between every single neuron. Hm. It would seem that we could have some groups focusing on neurons, another on types of neurons, another on dendritic tree structures, some more on the abstractions of dendritic trees, etc. in an up-*and*-down propagation hierarchy so that the abstract processes of the brain are studied just as well as the in-betweens of brain architecture. I was really thinking of the data collection problem: we cannot take one brain and get full information about all those things, down to a sufficient level of detail. I do not see such a technology even over the horizon (short of full-blow nanotechnology) that can deliver that. We can get different information from different individual brains (all of them dead), but combining that would not necessarily be meaningful: all brains are different. I think that if they did the whole project at that level of detail it would amount to a possibly interesting hint at some of the wiring, of peripheral interest to people doing work at the cognitive system level. But that is all. You see no more possible value of such a project? Well, I think that it will have more value one day, but at such a late stage in the history of cognitive system building that it will essentially just be a mopping up operation. In other words, we will have to do so much work at the cognitive level to be able to make sense of the wiring diagrams, that by that stage we will be able to generate our own systems. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65002389-10cd4a
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
RL:In order to completely ground the system, you need to let the system build its own symbols V. much agree with your whole argument. But - I may well have missed some vital posts - I have yet to get the slightest inkling of how you yourself propose to do this. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65013351-96e8f0
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
On Nov 14, 2007 1:36 PM, Mike Tintner [EMAIL PROTECTED] wrote: RL:In order to completely ground the system, you need to let the system build its own symbols Correct. Novamente is designed to be able to build its own symbols. what is built-in, are mechanisms for building symbols, and for probabilistically interrelating symbols once created... ben g V. much agree with your whole argument. But - I may well have missed some vital posts - I have yet to get the slightest inkling of how you yourself propose to do this. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65100803-21ddd3
Re: [agi] What best evidence for fast AI?
On Wednesday 14 November 2007 11:55, Richard Loosemore wrote: I was really thinking of the data collection problem: we cannot take one brain and get full information about all those things, down to a sufficient level of detail. I do not see such a technology even over the horizon (short of full-blow nanotechnology) that can deliver that. We can get different information from different individual brains (all of them dead), but combining that would not necessarily be meaningful: all brains are different. Re: all brains are different. What about the possibilities of cloning mice and then proceeding to raise them in Skinner boxes with the exact same environmental conditions, the same stimulation routines, etc. ? Ideally this will give us a baseline mouse that is not only genetically similar, but also behaviorally similar to some degree. This would undoubtedly be helpful in this quest. - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65191157-9f3b24
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
On Wednesday 14 November 2007 11:28, Richard Loosemore wrote: The complaint is not your symbols are not connected to experience. Everyone and their mother has an AI system that could be connected to real world input. The simple act of connecting to the real world is NOT the core problem. Are we sure? How much of the real world are we able to get into our AGI models anyway? Bandwidth is limited, much more limited than in humans and other animals. In fact, it might be the equivalent to worm tech. To do the calculations would I just have to check out how many neurons are in a worm, how many sensory neurons, and rough information theoretic estimations as to the minimum and maximums as to amounts of information processing that the worm's sensorium could be doing? - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65191610-b12544
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Bryan Bishop wrote: On Wednesday 14 November 2007 11:28, Richard Loosemore wrote: The complaint is not your symbols are not connected to experience. Everyone and their mother has an AI system that could be connected to real world input. The simple act of connecting to the real world is NOT the core problem. Are we sure? How much of the real world are we able to get into our AGI models anyway? Bandwidth is limited, much more limited than in humans and other animals. In fact, it might be the equivalent to worm tech. To do the calculations would I just have to check out how many neurons are in a worm, how many sensory neurons, and rough information theoretic estimations as to the minimum and maximums as to amounts of information processing that the worm's sensorium could be doing? I'm not quite sure where this is at . but the context of this particular discussion is the notion of 'symbol grounding' raised by Steven Harnad. I am essentially talking about how to solve the problem he described, and what exactly the problem was. Hence a lot of background behind this one, which if you don't know it might make it confusing. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65202116-6cf6d0
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
On Nov 14, 2007 11:58 PM, Bryan Bishop [EMAIL PROTECTED] wrote: Are we sure? How much of the real world are we able to get into our AGI models anyway? Bandwidth is limited, much more limited than in humans and other animals. In fact, it might be the equivalent to worm tech. To do the calculations would I just have to check out how many neurons are in a worm, how many sensory neurons, and rough information theoretic estimations as to the minimum and maximums as to amounts of information processing that the worm's sensorium could be doing? Pretty much. Let's take as our reference computer system a bog standard video camera connected to a high-end PC, which can do something (video compression, object recognition or whatever) with the input in real time. On the worm side, consider the model organism Caenorhabditis elegans, which has a few hundred neurons. It turns out that the computer has much more bandwidth. Then again, while intelligence unlike bandwidth isn't a scalar quantity even to a first approximation, to the extent they are comparable our best computer systems do seem to be considerably smarter than C. elegans. If we move up to something like a mouse, then the mouse has intelligence we can't replicate, and also has much more bandwidth than the computer system. Insects are somewhere in between, enough so that the comparison (both bandwidth and intelligence) doesn't produce an obvious answer; it's therefore considered not unreasonable to say present-day computers are in the ballpark of insect-smart. Of course that doesn't mean if we took today's software and connected it to mouse-bandwidth hardware it would become mouse-smart, but hopefully it means when we have that hardware we'll be able to use it to develop software that matches some of the things mice can do. (And it's still my opinion that by accepting - embracing - slowness on existing hardware we can work on the software at the same time as the hardware guys are working on their end, parallel rather than serial development.) - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65207531-031731
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Sounds a little confusing. Sounds like you plan to evolve a system through testing thousands of candidate mechanisms. So one way or another you too are taking a view - even if it's an evolutionary, I'm not taking a view view - on, and making a lot of asssumptions about -how systems evolve -the known architecture of human cognition. about which science has extremely patchy and confused knowledge. I don't see how any system-builder can avoid taking a view of some kind on such matters, yet you seem to be criticising Ben for so doing. I was hoping that you also had some view on how a system 's symbols should be grounded, especially since you mention Harnad, who does make vague gestures towards the brain's levels of grounding. But you don't indicate any such view. Sounds like you too, pace MW, are hoping for a number of miracles - IOW creative ideas - to emerge, and make your system work. Anyway, you have to give Ben credit for putting a lot of his stuff principles out there on the line. I think anyone who wants to mount a full-scale assault on him ( why not?) should be prepared to reciprocate. - RL: Mike Tintner wrote: RL:In order to completely ground the system, you need to let the system build its own symbols V. much agree with your whole argument. But - I may well have missed some vital posts - I have yet to get the slightest inkling of how you yourself propose to do this. Well, for the purposes of the present discussion I do not need to say how, only to say that there is a difference between two different research strategies for finding out what the mechanism is that does this. One strategy (the one that I claim has serious problems) is where you try to have your cake and eat it too: let the system build its own symbols, with attached parameters that 'mean' whatever they end up meaning after the symbols have been built, BUT then at the same time insist that some of the parameters really do 'mean' things like probabilities or likelihood or confidence values. If the programmer does anything at all to include mechanisms that rely on these meanings (these interpretations of what the parameters signify) then the programmer has second-guessed what the system itself was going to use those things for, and you have a conflict between the two. My strategy is to keep my hands off, not do anything to strictly interpret those parameters, and experimentally observe the properties of systems that seem loosely consistent with the known architecture of human cognition. I have a parameter, for instance, that seems to be a happiness or consistency parameter attached to a knowledge-atom. But beyond roughly characterising it as such, I do not insert any mechanisms that (implicitly or explicitly) lock the system into such an intepretation. Instead, I have a wide variety of different candidate mechanisms that use that parameter, and I look at the overall properties of systems that use these different candidate mechanisms. I let the system use the parameter according to the dictates of whatever mechanism is in place, but then I just explore the consequences (the high level behavior of the system). In this way I do not get a conflict between what I think the parameter 'ought' to mean and what the system is implicitly taking it to 'mean' by its use of the parameter. I could start talking about all the different candidate mechanisms, but there are thousands of them (at least thousands of candidates that I go so far as to test: they are generated in a semi-automatic way, so there are an unlimited number of potential candidates). Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.503 / Virus Database: 269.15.30/1125 - Release Date: 11/11/2007 9:50 PM - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65232546-91c089
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Mike Tintner wrote: RL:In order to completely ground the system, you need to let the system build its own symbols V. much agree with your whole argument. But - I may well have missed some vital posts - I have yet to get the slightest inkling of how you yourself propose to do this. Well, for the purposes of the present discussion I do not need to say how, only to say that there is a difference between two different research strategies for finding out what the mechanism is that does this. One strategy (the one that I claim has serious problems) is where you try to have your cake and eat it too: let the system build its own symbols, with attached parameters that 'mean' whatever they end up meaning after the symbols have been built, BUT then at the same time insist that some of the parameters really do 'mean' things like probabilities or likelihood or confidence values. If the programmer does anything at all to include mechanisms that rely on these meanings (these interpretations of what the parameters signify) then the programmer has second-guessed what the system itself was going to use those things for, and you have a conflict between the two. My strategy is to keep my hands off, not do anything to strictly interpret those parameters, and experimentally observe the properties of systems that seem loosely consistent with the known architecture of human cognition. I have a parameter, for instance, that seems to be a happiness or consistency parameter attached to a knowledge-atom. But beyond roughly characterising it as such, I do not insert any mechanisms that (implicitly or explicitly) lock the system into such an intepretation. Instead, I have a wide variety of different candidate mechanisms that use that parameter, and I look at the overall properties of systems that use these different candidate mechanisms. I let the system use the parameter according to the dictates of whatever mechanism is in place, but then I just explore the consequences (the high level behavior of the system). In this way I do not get a conflict between what I think the parameter 'ought' to mean and what the system is implicitly taking it to 'mean' by its use of the parameter. I could start talking about all the different candidate mechanisms, but there are thousands of them (at least thousands of candidates that I go so far as to test: they are generated in a semi-automatic way, so there are an unlimited number of potential candidates). Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65198894-3ece99
Re: [agi] What best evidence for fast AI?
Bryan Bishop wrote: On Wednesday 14 November 2007 11:55, Richard Loosemore wrote: I was really thinking of the data collection problem: we cannot take one brain and get full information about all those things, down to a sufficient level of detail. I do not see such a technology even over the horizon (short of full-blow nanotechnology) that can deliver that. We can get different information from different individual brains (all of them dead), but combining that would not necessarily be meaningful: all brains are different. Re: all brains are different. What about the possibilities of cloning mice and then proceeding to raise them in Skinner boxes with the exact same environmental conditions, the same stimulation routines, etc. ? Ideally this will give us a baseline mouse that is not only genetically similar, but also behaviorally similar to some degree. This would undoubtedly be helpful in this quest. Well, now you have suggested this I am sure some neuroscientist will do it ;-). But you have to understand that I am a cognitive scientist, with a huge agenda that involves making good use of what I see as the uneplxored fertile ground between cognitive science and AI and I think that I will be able to build an AGI using this approach *long* before the neuroscientists even get one mouse-brain scan at the neuron level (never mind the synaptic bouton level)! So: yeah, but not necessary. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=65204588-4868d1
Re: [agi] What best evidence for fast AI?
Bryan Bishop wrote: On Monday 12 November 2007 22:16, Richard Loosemore wrote: If anyone were to throw that quantity of resources at the AGI problem (recruiting all of the planet), heck, I could get it done in about 3 years. ;-) I have done some research on this topic in the last hour and have found that a Connectome Project is in fact in the very early stages out there on the internet: http://iic.harvard.edu/projects/connectome.html http://acenetica.blogspot.com/2005/11/human-connectome.html http://acenetica.blogspot.com/2005/10/mission-to-build-simulated-brain.html http://www.indiana.edu/~cortex/connectome_plos.pdf This is the whole brain emulation approach, I guess (my previous comments were about evolution of brains rather than neural level duplication). But (switching topics to whole brain emulation) there are serious problems with this. It seems quite possible that what we need is a detailed map of every synapse, exact layout of dendritic tree structures, detailed knowledge of the dynamics of these things (they change rapidly) AND wiring between every single neuron. When I say it seems possible I mean that the chance of this information being absolutely necessary in order to understand what the neural system is doing, is so high that we would not want to gamble on them NOT being necessary. So are the researchers working at that level of detail? Egads, no! Here's a quote from the PLOS Computational Biology paper you referenced (above): Attempting to assemble the human connectome at the level of single neurons is unrealistic and will remain infeasible at least in the near future. They are not even going to do it at the resolution needed to see individual neurons?! I think that if they did the whole project at that level of detail it would amount to a possibly interesting hint at some of the wiring, of peripheral interest to people doing work at the cognitive system level. But that is all. I think it would be roughly equivalent to the following: You say to me I want to understand how computers work, in enough detail to build my own and I reply with I can get a you a photo of a motherboard and a 500 by 500 pixel image of the inside of an Intel chip... Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64523531-24742d
Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
flexibility is what compounds the problem. Remember, life exists on the boundary between order and chaos. Too much flexibility (unconstrained chaos) is as deadly as too much structure. I think that I see both sides of the issue and how Novamente could be altered/enhanced to make Richard happy (since it's almost universally flexible) -- but doing so would also impose many constraints that I think that you would be unwilling to live with since I'm not sure that you would see the point. I don't think that you're ever going to be able to change his view that the current direction of Novamente is -- pick one: a) a needle in an infinite haystack or b) too fragile to succeed -- particularly since I'm pretty sure that you couldn't convince me without making some serious additions to Novamente - Original Message - *From:* Benjamin Goertzel mailto:[EMAIL PROTECTED] *To:* agi@v2.listbox.com mailto:agi@v2.listbox.com *Sent:* Monday, November 12, 2007 3:49 PM *Subject:* Re: [agi] What best evidence for fast AI? To be honest, Richard, I do wonder whether a sufficiently in-depth conversation about AGI between us would result in you changing your views about the CSP problem in a way that would accept the possibility of Novamente-type solutions. But, this conversation as I'm envisioning it would take dozens of hours, and would require you to first spend 100+ hours studying detailed NM materials, so this seems unlikely to happen in the near future. -- Ben On Nov 12, 2007 3:32 PM, Richard Loosemore [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote: Benjamin Goertzel wrote: Ed -- Just a quick comment: Mark actually read a bunch of the proprietary, NDA-required Novamente documents and looked at some source code (3 years ago, so a lot of progress has happened since then). Richard didn't, so he doesn't have the same basis of knowledge to form detailed comments on NM, that Mark does. This is true, but not important to my line of argument, since of course I believe that a problem exists (CSP), which we have discussed on a number of occasions, and your position is not that you have some proprietary, unknown-to-me solution to the problem, but rather that you do not really think there is a problem. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; http://v2.listbox.com/member/?; This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; http://v2.listbox.com/member/?; This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64591405-3ee8b5
RE: [agi] What best evidence for fast AI?
probabilistic formulas are appropriate. But it doesnt argue against the importance of probabilities It argues against using them blindly. ED So by operating with small amounts of data how small, very roughly, are you talking about. And are you only talking about the active goals or sources of activation, that will be small or are you saying that all the computation in the system will only be dealing with a small amount of data within, for example, one second of the processing of human-level system operating at human-level speed? MARK I mean like the way humans reason, there is only concentration on a small number of objects -- which are only one link away from an almost inconceivable number of related things -- and then the brain can jump at least three of these links with lightning rapidity. ED So this implies you are not arguing against the idea that AGI will be dealing with massive data, just that that use will be focused by a concentration on a relatively small number of sources of activation at once. MARK Ask Ben how much actual work has been done on activation control in very large, very sparse atom spaces in Novamente. He'll tell you that it's a project for when he's further along. I'll insist (as will Richard) that if it isn't baked in from the very beginning, you're probably going to have to go back to the beginning to repair the lack. ED It is exactly such research I want to see funded. It strikes me as one of the key things we must learn to do well to make powerful AGI. But I think even with some fairly dumb activation control systems you could get useful results. Such results would not be at all human-level in may ways, but in other ways they could be much more powerful because such systems could deal with many more explicit facts and could input and output information at a much higher rate than humans. For example, what is the equivalent of the activation control (or search) algorithm in Google sets. They operate over huge data. I bet the algorithm for calculating their search or activation is relatively simple (much, much, much less than a PhD theses) and look what they can do. So I think one path is to come up with applications that can use and reason with large data, having roughly world knowledge-like sparseness, (such as NL data) and start with relatively simple activation algorithms and develop then from the ground up. MARK P.S. Oh yeah -- if you were public enemy number one, I wouldn't bother answering you (and I probably should lay off of the fan-boy crap :-). ED Thanks. I admit I am impressed with Novamente. Since its the best AGI architecture I currently know of; I am impressed with Ben; believe there is a high probability all the gaps you address could be largely fixed within five years with deep funding (which may never come); and since I want to get such deep funding for just the type of large atom-base work you say is so critical, I think it is important to focus on the potential for greatness that Novamente and somewhat similar systems have, rather than only think of its current gaps and potential problems. But of course, at the same time, we must look for and try to understand its gaps and potential problems so that we can remove them. Ed Porter -Original Message- From: Mark Waser [mailto:[EMAIL PROTECTED] Sent: Monday, November 12, 2007 2:42 PM To: agi@v2.listbox.com Subject: Re: [agi] What best evidence for fast AI? It is NOT clear that Novamente documentation is NOT enabling, or could not be made enabling, with, say, one man year of work. Strong argument could be made both ways. I believe that Ben would argue that Novamente documentation is NOT enabling even with one man-year of work. Ben? There is still way to much *research* work to be done. But the standard for non-enablement is very arguably weaker than not requiring a miracle. It would be more like not requiring a leap of creativity that is outside the normal skill of talented PhDs trained in related fields. So although your position is reasonable, I hope you understand so is that on the other side. My meant-to-be-humorous miracle phrasing is clearly throwing you. The phrase not requiring a leap of creativity that is outside the normal skill of talented PhDs trained in related fields works for me. Novamente is *definitely* not there yet. I'm rather sure that Ben would agree -- as in, I'm not on the other side, *you* are on the other side from the system's designer. Again, Ben please feel free to chime in. much scaling stuff Remember that the brain is *massively* parallel. Novamente and any other linear (or minorly-parallel) system is *not* going to work in the same fashion as the brain. Novamente can be parallelized to some degree but *not* to anywhere near the same degree as the brain. I love your speculation and agree with it -- but it doesn't match near-term reality. We aren't going to have brain-equivalent parallelism
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
view that the current direction of Novamente is -- pick one: a) a needle in an infinite haystack or b) too fragile to succeed -- particularly since I'm pretty sure that you couldn't convince me without making some serious additions to Novamente - Original Message - *From:* Benjamin Goertzel mailto:[EMAIL PROTECTED] *To:* agi@v2.listbox.com mailto:agi@v2.listbox.com *Sent:* Monday, November 12, 2007 3:49 PM *Subject:* Re: [agi] What best evidence for fast AI? To be honest, Richard, I do wonder whether a sufficiently in-depth conversation about AGI between us would result in you changing your views about the CSP problem in a way that would accept the possibility of Novamente-type solutions. But, this conversation as I'm envisioning it would take dozens of hours, and would require you to first spend 100+ hours studying detailed NM materials, so this seems unlikely to happen in the near future. -- Ben On Nov 12, 2007 3:32 PM, Richard Loosemore [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote: Benjamin Goertzel wrote: Ed -- Just a quick comment: Mark actually read a bunch of the proprietary, NDA-required Novamente documents and looked at some source code (3 years ago, so a lot of progress has happened since then). Richard didn't, so he doesn't have the same basis of knowledge to form detailed comments on NM, that Mark does. This is true, but not important to my line of argument, since of course I believe that a problem exists (CSP), which we have discussed on a number of occasions, and your position is not that you have some proprietary, unknown-to-me solution to the problem, but rather that you do not really think there is a problem. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; http://v2.listbox.com/member/?; This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; http://v2.listbox.com/member/?; This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64606349-2f1f37
Re: [agi] What best evidence for fast AI?
For example, what is the equivalent of the activation control (or search) algorithm in Google sets. They operate over huge data. I bet the algorithm for calculating their search or activation is relatively simple (much, much, much less than a PhD theses) and look what they can do. So I think one path is to come up with applications that can use and reason with large data, having roughly world knowledge-like sparseness, (such as NL data) and start with relatively simple activation algorithms and develop then from the ground up. Google, I believe, does reasoning about word and phrase co-occurrence using a combination of Bayes net learning with EM clustering (this is based on personal conversations with folks who have worked on related software there). The use of EM helps the Bayes net approach scale. Bayes nets are good for domains like word co-occurence probabilities, in which the relevant data is relatively static. They are not much good for real-time learning. Unlike Bayes nets, the approach taken in PLN and NARS allows efficient uncertain reasoning in dynamic environments based on large knowledge bases (at least in principle, based on the math, algorithms and structures; we haven't proved it yet). -- Ben G - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64609544-b69ea5
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Mike Tintner wrote: RL:Suppose that in some significant part of Novamente there is a representation system that uses probability or likelihood numbers to encode the strength of facts, as in [I like cats](p=0.75). The (p=0.75) is supposed to express the idea that the statement [I like cats] is in some sense 75% true. This essay seems to be a v.g. demonstration of why the human system almost certainly does not use numbers or anything like, as stores of value - but raw, crude emotions. How much do you like cats [or marshmallow ice cream]? Miaow//[or yummy] [those being an expression of internal nervous and muscular impulses] And black cats [or strawberry marshmallow] ? Miaow-miaoww![or yummy yummy] . It's crude but it's practical. It is all a question of what role the numbers play. Conventional AI wants them at the surface, and transparently interpretable. I am not saying that there are no numbers, but only that they are below the surface, and not directly interpretable. that might or might not gibe with what you are saying ... although I would not go so far as to put it in the way you do. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64636829-14d428
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
Ben, Unfortunately what you say below is tangential to my point, which is what happens when you reach the stage where you cannot allow any more vagueness or subjective interpretation of the qualifiers, because you have to force the system to do its own grounding, and hence its own interpretation. What you gave below was a sketch of some more elaborate 'qualifier' mechanisms. But I described the process of generating more and more elaborate qualifier mechanisms in the body of the essay, and said why this process was of no help in resolving the issue. Richard Loosemore Benjamin Goertzel wrote: Richard, The idea of the PLN semantics underlying Novamente's probabilistic truth values is that we can have **both** -- simple probabilistic truth values without highly specific interpretation -- more complex, logically refined truth values, when this level of precision is necessary To make the discussion more concrete, I'll use a specfic example to do with virtual animals in Second Life. Our first version of the virtual pets won't use PLN in this sort of way, it'll be focused on MOSES evolutionary learning; but, this is planned for the second version and is within the scope of what Novamente can feasibly be expected to do with modest effort. Consider an avatar identified as Bob_Yifu And, consider the concept of friend, which is a ConceptNode -- associated to the WordNode friend via a learned ReferenceLink -- defined operationally via a number of links such as ImplicationLink AND InheritanceLink X friend EvaluationLink near (I, X) Pleasure (this one just says that being near a friend confers pleasure. Other links about friendship may contain knowledge such as that friends often give one food, friends help one find things, etc.) The concept of friend may be learned, via mining of the animal's experience-base -- basically, this is a matter of learning that there are certain predicates whose SatisfyingSets (the set of Atoms that fulfill the predicate) have significant intersection, and creating a ConceptNode to denote that intersection. Then, once the concept of friend has been formed, more links pertaining to it may be learned via mining the experience base and via inference rules. Then, we can may find that InheritanceLink Bob_Yifu friend .9,1 (where the .9,1 is an interval probability, interpreted according to the indefinite probabilities framework) and this link mixes intensional and extensional inheritance, and thus is only useful for heuristic reasoning (which however is a very important kind). What this link means is basically that Bob_Yifu's node in the memory has a lot of the same links as the friend node -- or rather, that it **would**, if all its links were allowed to exist rather than being pruned to save memory. So, note that the semantics are actually tied to the mind itself. Or we can make more specialized logical constructs if we really want to, denoting stuff like -- at certain times Bob_Yifu is a friend -- Bob displays some characteristics of friendship very strongly, and others not at all -- etc. We can also do crude, heuristic contextualization like ContextLink .7,.8 home InheritanceLink Bob_Yifu friend which suggests that Bob is less friendly at home than in general. Again this doesn't capture all the subtleties of Bob's friendship in relation to being at home -- and one could do so if one wanted to, but it would require introducing a larger complex of nodes and links, which is not always the most appropriate thing to do. The PLN inference rules are designed to give heuristically correct conclusions based on heuristically interpreted links; or more precise conclusions based on more precisely interpreted links. Finally, the semantics of PLN relationships is explicitly an **experiential** semantics. (One of the early chapters in the PLN book, to appear via Springer next year, is titled Experiential Semantics.) So, all node and link truth values in PLN are intended to be settable and adjustable via experience, rather than via programming or importation from databases or something like that. Now, the above example is of course a quite simple one. Discussing a more complex example would go beyond the scope of what I'm willing to do in an email conversation, but the mechanisms I've described are not limited to such simple examples. I am aware that identifying Bob_Yifu as a coherent, distinct entity is a problem faced by humans and robots, and eliminated via the simplicity of the SL environment. However, there is detailed discussion in the (proprietary) NM book of how these same mechanisms may be used to do object recognition and classification, as well. You may of course argue that these mechanisms won't scale up to large knowledge bases and rich experience streams. I believe that they will, and have arguments but not rigorous proofs that they will. -- Ben G On Nov 13, 2007 12:34 PM, Richard Loosemore
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
On Nov 13, 2007 2:37 PM, Richard Loosemore [EMAIL PROTECTED] wrote: Ben, Unfortunately what you say below is tangential to my point, which is what happens when you reach the stage where you cannot allow any more vagueness or subjective interpretation of the qualifiers, because you have to force the system to do its own grounding, and hence its own interpretation. I don't see why you talk about forcing the system to do its own grounding -- the probabilities in the system are grounded in the first place, as they are calculated based on experience. The system observes, records what it sees, abstracts from it, and chooses actions that it guess will fulfill its goals. Its goals are ultimately grounded in in-built feeling-evaluation routines, measuring stuff like amount of novelty observed, amount of food in system etc. So, the system sees and then acts ... and the concepts it forms and uses are created/used based on their utility in deriving appropriate actions. There is no symbol-grounding problem except in the minds of people who are trying to interpret what the system does, and get confused. Any symbol used within the system, and any probability calculated by the system, are directly grounded in the system's experience. There is nothing vague about an observation like Bob_Yifu was observed at time-stamp 599933322, or a fact Command 'wiggle ear' was sent at time-stamp 54. These perceptions and actions are the root of the probabilities the system calculated, and need no further grounding. What you gave below was a sketch of some more elaborate 'qualifier' mechanisms. But I described the process of generating more and more elaborate qualifier mechanisms in the body of the essay, and said why this process was of no help in resolving the issue. So, if a system can achieve its goals based on choosing procedures that it thinks are likely to achieve its goals, based on the knowledge it gathered via its perceived experience -- why do you think it has a problem? I don't really understand your point, I guess. I thought I did -- I thought your point was that precisely specifying the nature of a conditional probability is a rats-nest of complexity. And my response was basically that in Novamente we don't need to do that, because we define conditional probabilities based on the system's own knowledge-base, i.e. Inheritance A B .8 means If A and B were reasoned about a lot, then A would (as measred by an weighted average) have 80% of the relationships that B does But apparently you were making some other point, which I did not grok, sorry... Anyway, though, Novamente does NOT require logical relations of escalating precision and complexity to carry out reasoning, which is one thing you seemed to be assuming in your post. Ben - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64644318-8bbdee
Re: [agi] What best evidence for fast AI?
On Mon, Nov 12, 2007 at 08:44:58PM -0500, Mark Waser wrote: So perhaps the AGI question is, what is the difference between a know-it-all mechano-librarian, and a sentient being? I wasn't assuming a mechano-librarian. I was assuming a human that could (and might be trained to) do some initial translation of the question and some final rephrasing of the answer. I'm surprised by your answer. I don't see that the hardest part of agi is NLP i/o. To put it into perspective: one can fake up some trivial NLP i/o now, and with a bit of effort, one can improve significantly on that. Sure, it would be child-like conversation, and the system would be incapable of learning new idioms, expressions, etc., but I don't see that you'd need a human to translate the question into some formal reasoning-engine language. The hard part of NLP is being able to read complex texts, whether Alexander Pope or Karl Marx; but a basic NLP i/o interface stapled to a reasoning engine doesn't need to really do that, or at least not well. Yet, these two stapled toegether would qualify as a mechano-librarian for me. To me, the hard part is still the reasoning engine itself, and the pruning, and the tailoring of responses to the topic at hand. So let me rephrase the question: If one had 1) A reasoing engine that could provide short yet appropriate responses to questions, 2) A simple NLP interface to the reasoning engine would that be AGI? I imagine most folks would say no, so let me throw in: 3) System can learn new NLP idioms, so that it can eventually come to understand those sentences and paragraphs that make Karl Marx so hard to read. With this enhanced reading ability, it could then presumably become a know-it-all ultra-question-answerer. Would that be AGI? Or is there yet more? Well, of course there's more: one expects creativity, aesthetics, ethics. But we know just about nothing about that. This is the thing that I think is relevent to Robin Hanson's original question. I think we can build 1+2 is short order, and maybe 3 in a while longer. But the result of 1+2+3 will almost surely be an idiot-savant: knows everything about horses, and can talk about them at length, but, like a pedantic lecturer, the droning will put you asleep. So is there more to AGI, and exactly how do way start laying hands on that? --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64661358-af169f
Re: [agi] What best evidence for fast AI?
This is the thing that I think is relevent to Robin Hanson's original question. I think we can build 1+2 is short order, and maybe 3 in a while longer. But the result of 1+2+3 will almost surely be an idiot-savant: knows everything about horses, and can talk about them at length, but, like a pedantic lecturer, the droning will put you asleep. So is there more to AGI, and exactly how do way start laying hands on that? --linas I think that evolutionary-learning-type methods play a big role in creativity. I elaborated on this quite a bit toward the end of my 1997 book From Complexity to Creativity. Put simply, inference is ultimately a local search method -- inference rules, even heuristic and speculative ones, always lead you step by step from what you know into the unknown. This makes you, as you say, like a pedantic lecturer. OTOH, evolutionary algorithms can take big creative leaps. This is one reason why the MOSES evolutionary algorithm plays a big role in the Novamente design (the other, related reason being that evolutionary learning is better than logical inference for many kinds of procedure learning). Integrating evolution with logic is key to intelligence. The brain does it, I believe, via -- implementing logic via Hebbian learning (neuron-level Hebb stuff leading to PLN-like logic stuff on the neural-assembly level) -- implementing evolution via Edelman-style Neural Darwinist neural map evolution (which ultimately bottoms out in Hebbian learning too) Novamente seeks to enable this integration via grounding both inference and evolutionary learning in probability theory. -- Ben G -- Ben G - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64667888-a48aa3
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
On Tue, Nov 13, 2007 at 12:34:51PM -0500, Richard Loosemore wrote: Suppose that in some significant part of Novamente there is a representation system that uses probability or likelihood numbers to encode the strength of facts, as in [I like cats](p=0.75). The (p=0.75) is supposed to express the idea that the statement [I like cats] is in some sense 75% true. Either way, we have a problem: a fact like [I like cats](p=0.75) is ungrounded because we have to interpret it. Does it mean that I like cats 75% of the time? That I like 75% of all cats? 75% of each cat? Are the cats that I like always the same ones, or is the chance of an individual cat being liked by me something that changes? Does it mean that I like all cats, but only 75% as much as I like my human family, which I like(p=1.0)? And so on and so on. Eh? You are standing at the proverbial office water coooler, and Aneesh says Wen likes cats. On your drive home, you mind races .. does this mean that Wen is a cat fancier? You were planning on taking Wen out on a date, and this tidbit of information could be useful ... when you try to build the entire grounding mechanism(s) you are forced to become explicit about what these numbers mean, during the process of building a grounding system that you can trust to be doing its job: you cannot create a mechanism that you *know* is constructing sensible p numbers and facts during all of its development *unless* you finally bite the bullet and say what the p numbers really mean, in fully cashed out terms. But has a human, asking Wen out on a date, I don't really know what Wen likes cats ever really meant. It neither prevents me from talking to Wen, or from telling my best buddy that ...well, I know, for instance, that she likes cats... Lack of grounding is what makes humour funny, you can do a whole Pygmalion / Seinfeld episode on she likes cats. --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64672202-2af80e
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
But has a human, asking Wen out on a date, I don't really know what Wen likes cats ever really meant. It neither prevents me from talking to Wen, or from telling my best buddy that ...well, I know, for instance, that she likes cats... yes, exactly... The NLP statement Wen likes cats is vague in the same way as the Novamente or NARS relationship EvaluationLink likes ListLink Wen cats is vague The vagueness passes straight from NLP into the internal KR, which is how it should be. And that same vagueness may be there if the relationship is learned via inference based on experience, rather than acquired by natural language. I.e., if the above relationship is inferred, it may just mean that {the relationship between Wen and cats} shares many relationships with other person/object relationships that have been categorized as 'liking' before In this case, the system can figure out that Wen likes cats without ever actually making explicit what this means. All it knows is that, whatever it means, it's the same thing that was meant in other circumstances where liking was used as a label. So, vagueness can not only be important into an AI system from natural language, but also propagated around the AI system via inference. This is NOT one of the trickier things about building probabilistic AGI, it's really kind of elementary... -- Ben G - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64674694-3ada83
Re: Essay - example of how the CSP bites [WAS Re: [agi] What best evidence for fast AI?]
So, vagueness can not only be important imported, I meant into an AI system from natural language, but also propagated around the AI system via inference. This is NOT one of the trickier things about building probabilistic AGI, it's really kind of elementary... -- Ben G - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64674943-4b25e0
Re: [agi] What best evidence for fast AI?
I don't see that the hardest part of agi is NLP i/o. I didn't say that i/o was the hardest part of agi. Truly understanding NLP is agi-complete though. And please, get off this kick of just faking something up and thinking that because you can create a shallow toy example that holds for ten seconds that you've answered *anything*. That's the *narrow ai* approach. - Original Message - From: Linas Vepstas [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Tuesday, November 13, 2007 4:01 PM Subject: Re: [agi] What best evidence for fast AI? On Mon, Nov 12, 2007 at 08:44:58PM -0500, Mark Waser wrote: So perhaps the AGI question is, what is the difference between a know-it-all mechano-librarian, and a sentient being? I wasn't assuming a mechano-librarian. I was assuming a human that could (and might be trained to) do some initial translation of the question and some final rephrasing of the answer. I'm surprised by your answer. I don't see that the hardest part of agi is NLP i/o. To put it into perspective: one can fake up some trivial NLP i/o now, and with a bit of effort, one can improve significantly on that. Sure, it would be child-like conversation, and the system would be incapable of learning new idioms, expressions, etc., but I don't see that you'd need a human to translate the question into some formal reasoning-engine language. The hard part of NLP is being able to read complex texts, whether Alexander Pope or Karl Marx; but a basic NLP i/o interface stapled to a reasoning engine doesn't need to really do that, or at least not well. Yet, these two stapled toegether would qualify as a mechano-librarian for me. To me, the hard part is still the reasoning engine itself, and the pruning, and the tailoring of responses to the topic at hand. So let me rephrase the question: If one had 1) A reasoing engine that could provide short yet appropriate responses to questions, 2) A simple NLP interface to the reasoning engine would that be AGI? I imagine most folks would say no, so let me throw in: 3) System can learn new NLP idioms, so that it can eventually come to understand those sentences and paragraphs that make Karl Marx so hard to read. With this enhanced reading ability, it could then presumably become a know-it-all ultra-question-answerer. Would that be AGI? Or is there yet more? Well, of course there's more: one expects creativity, aesthetics, ethics. But we know just about nothing about that. This is the thing that I think is relevent to Robin Hanson's original question. I think we can build 1+2 is short order, and maybe 3 in a while longer. But the result of 1+2+3 will almost surely be an idiot-savant: knows everything about horses, and can talk about them at length, but, like a pedantic lecturer, the droning will put you asleep. So is there more to AGI, and exactly how do way start laying hands on that? --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64683060-82d4be
Re: [agi] What best evidence for fast AI?
On Tuesday 13 November 2007 09:11, Richard Loosemore wrote: This is the whole brain emulation approach, I guess (my previous comments were about evolution of brains rather than neural level duplication). Ah, you are right. But this too is an interesting topic. I think that the order of magnitudes for whole brain emulation, connectome, and similar evolutionary methods, are roughly the same, but I haven't done any calculations. It seems quite possible that what we need is a detailed map of every synapse, exact layout of dendritic tree structures, detailed knowledge of the dynamics of these things (they change rapidly) AND wiring between every single neuron. Hm. It would seem that we could have some groups focusing on neurons, another on types of neurons, another on dendritic tree structures, some more on the abstractions of dendritic trees, etc. in an up-*and*-down propagation hierarchy so that the abstract processes of the brain are studied just as well as the in-betweens of brain architecture. I think that if they did the whole project at that level of detail it would amount to a possibly interesting hint at some of the wiring, of peripheral interest to people doing work at the cognitive system level. But that is all. You see no more possible value of such a project? - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64757679-f3c1ec
Re: [agi] What best evidence for fast AI?
I was using the term episodic in the standard sense of episodic memory from cog psych, in which episodic memory is differentiated from procedural and declarative memory. I understood that. The problem is that procedural and declarative memory is *not* as simple as is often purported. If you can't rapidly realize when and why your previously reliable procedural and declarative stuff is suddenly no longer valid . . . . The main point is, we have specialized indices to make memory access efficient for knowledge involving (certain and uncertain) logical relationships, associations, spatial and temporal relationships, and procedures Indices are important but compactness of data storage is also important as are ways to have what is effectively indexed derivation of knowledge. Obviously my knowledge of Novamente is becoming dated but, unless you opened some really new areas, there is a lot of work that could be done in this area that you're not focusing on. (Note: Please don't be silly infer that by compactness of data storage that I mean that disk size is important -- we're long past those days. Assume that I mean the computational costs of manipulating data that is not stored in an efficient manner). Research project 1. How do you find analogies between neural networks, enzyme kinetics and the formation of galaxies (hint: think Boltzmann)? That is a question most humans couldn't answer, and is only suitable for testing an AGI that is already very advanced. In your opinion. I don't believe that an AGI is going to get far at all without having at least a partial handle on this. Research project 2. How do you recognize and package up all of the data that represents horse and expose only that which is useful at a given time? That is covered quite adequately in the NM design, IMO. We are actually doing a commercial project right now (w/ delivery in 2008) that will showcase our ability to solve this problem. Details are confidential unfortunately, due to the customer's preference. I'm afraid that I have to snort at this. Either you didn't understand the full implications of what I'm saying or you're snowing me (ok, I'll give you a .1% chance of having it). That is what is called map encapsulation in the Novamente design. Yes, yes, I saw it in the design . . . . a miracle happens here. Which, granted, is better than not realizes that the area exists . . . . but still . . . . I do not think the design has any huge gaps. But much further RD work is required, and I agree there may be a simpler approach; but I am not convinced that you have one. These are two *very* different issues (with a really spurious statement tacked onto the end). Of course you don't think the design has any gaps -- you would have filled them if you saw them. There is no reason to be convinced that *I* have a simpler approach because I haven't put one forth. I may or may not be working on one:-) but if I am, I certainly haven't got to the point where I feel that I can defend it.:-) - Original Message - From: Benjamin Goertzel To: agi@v2.listbox.com Sent: Monday, November 12, 2007 11:45 AM Subject: Re: [agi] What best evidence for fast AI? On Nov 12, 2007 11:36 AM, Mark Waser [EMAIL PROTECTED] wrote: I am extremely confident of Novamente's memory design regarding declarative and procedural knowledge. Tweaking the system for optimal representation of episodic knowledge may require some more thought. Granted -- the memory design is very generic and will handle virtually anything. The question is -- is it in a reasonably optimal from for retrieval and other operations (i.e. optimal enough that it won't end up being impossibly slow once you get a realistic amount of data/knowledge). Your caveat on episodic knowledge proves very informative since *all* knowledge is effectively episodic. I was using the term episodic in the standard sense of episodic memory from cog psych, in which episodic memory is differentiated from procedural and declarative memory. The main point is, we have specialized indices to make memory access efficient for knowledge involving (certain and uncertain) logical relationships, associations, spatial and temporal relationships, and procedures ... but we haven't put much work into creating specialized indices to make access of stories/narratives efficient. Though this may not wind up being necessary since the AtomTable now has the capability to create new indices on the fly, based on the statistics of the data contained therein. I have no idea what you mean by scale invariance of knowledge nor and only weak understanding of what you mean by ways of determining and exploiting encapsulation and modularity of knowledge without killing useful leaky abstractions. Research project 1. How do you find analogies between neural networks, enzyme
Re: [agi] What best evidence for fast AI?
Hi, Research project 1. How do you find analogies between neural networks, enzyme kinetics and the formation of galaxies (hint: think Boltzmann)? That is a question most humans couldn't answer, and is only suitable for testing an AGI that is already very advanced. In your opinion. I don't believe that an AGI is going to get far at all without having at least a partial handle on this. I'm more interested at this stage in analogies like -- btw seeking food and seeking understanding -- between getting an object out of a hole and getting an object out of a pocket, or a guarded room etc. Why would one need to introduce advanced scientific concepts to an early-stage AGI? I don't get it... Research project 2. How do you recognize and package up all of the data that represents horse and expose only that which is useful at a given time? That is covered quite adequately in the NM design, IMO. We are actually doing a commercial project right now (w/ delivery in 2008) that will showcase our ability to solve this problem. Details are confidential unfortunately, due to the customer's preference. I'm afraid that I have to snort at this. Either you didn't understand the full implications of what I'm saying or you're snowing me (ok, I'll give you a .1% chance of having it). Hmmm I guess I didn't understand what you meant. What I thought you meant was, if a user asked I'm a small farmer in New Zealand. Tell me about horses then the system would be able to disburse its relevant knowledge about horses, filtering out the irrelevant stuff. What did you mean, exactly? That is what is called map encapsulation in the Novamente design. Yes, yes, I saw it in the design . . . . a miracle happens here. Which, granted, is better than not realizes that the area exists . . . . but still . . . . There are specific algorithms proposed, in the NM book, for doing map encapsulation. You may not believe they will work for the task, but still, it's not fair to use the label a miracle happens here to describe a description of specific algorithms applied to a specific data structure. I do not think the design has any huge gaps. But much further RD work is required, and I agree there may be a simpler approach; but I am not convinced that you have one. These are two *very* different issues (with a really spurious statement tacked onto the end). Of course you don't think the design has any gaps -- you would have filled them if you saw them. I think it has medium-sized gaps, not huge ones. I have not filled all these gaps because of lack of time -- implementing stuff needs to be balanced with finalizing design details of stuff that won't be implemented for a while anyway due to limited resources. -- Ben - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64208314-2be377
Re: [agi] What best evidence for fast AI?
I'm more interested at this stage in analogies like -- btw seeking food and seeking understanding -- between getting an object out of a hole and getting an object out of a pocket, or a guarded room Why would one need to introduce advanced scientific concepts to an early-stage AGI? I don't get it... :-) A bit disingenuous there, Ben. Obviously you start with the simple and move on to the complex (though I suspect that the first analogy you cite is rather more complex than you might think) -- but to take too simplistic an approach that might not grow is just the narrow AI approach in other clothing. Hmmm I guess I didn't understand what you meant. What I thought you meant was, if a user asked I'm a small farmer in New Zealand. Tell me about horses then the system would be able to disburse its relevant knowledge about horses, filtering out the irrelevant stuff. What did you mean, exactly? That's a good simple, starting case. But how do you decide how much knowledge to disburse? How do you know what is irrelevant? How much do your answers differ between a small farmer in New Zealand, a rodeo rider in the West, a veterinarian is Pennsylvania, a child in Washington, a bio-mechanician studying gait? And horse is actually a *really* simple concept since it refers to a very specific type of physical object. Besides, are you really claiming that you'll be able to do this next year? Sorry, but that is just plain, unadulterated BS. If you can do that, you are light-years further along than . . . . There are specific algorithms proposed, in the NM book, for doing map encapsulation. You may not believe they will work for the task, but still, it's not fair to use the label a miracle happens here to describe a description of specific algorithms applied to a specific data structure. I guess that the jury will have to be out until you publicize the algorithms. What I've seen in the past are too small, too simple, and won't scale to what is likely to be necessary. I think it has medium-sized gaps, not huge ones. I have not filled all these gaps because of lack of time -- implementing stuff needs to be balanced with finalizing design details of stuff that won't be implemented for a while anyway due to limited resources. :-) You have more than enough design experience to know that medium-size gaps can frequently turn huge once you turn your attention to them. Who are you snowing here? - Original Message - From: Benjamin Goertzel To: agi@v2.listbox.com Sent: Monday, November 12, 2007 12:55 PM Subject: Re: [agi] What best evidence for fast AI? Hi, Research project 1. How do you find analogies between neural networks, enzyme kinetics and the formation of galaxies (hint: think Boltzmann)? That is a question most humans couldn't answer, and is only suitable for testing an AGI that is already very advanced. In your opinion. I don't believe that an AGI is going to get far at all without having at least a partial handle on this. I'm more interested at this stage in analogies like -- btw seeking food and seeking understanding -- between getting an object out of a hole and getting an object out of a pocket, or a guarded room etc. Why would one need to introduce advanced scientific concepts to an early-stage AGI? I don't get it... Research project 2. How do you recognize and package up all of the data that represents horse and expose only that which is useful at a given time? That is covered quite adequately in the NM design, IMO. We are actually doing a commercial project right now (w/ delivery in 2008) that will showcase our ability to solve this problem. Details are confidential unfortunately, due to the customer's preference. I'm afraid that I have to snort at this. Either you didn't understand the full implications of what I'm saying or you're snowing me (ok, I'll give you a .1% chance of having it). Hmmm I guess I didn't understand what you meant. What I thought you meant was, if a user asked I'm a small farmer in New Zealand. Tell me about horses then the system would be able to disburse its relevant knowledge about horses, filtering out the irrelevant stuff. What did you mean, exactly? That is what is called map encapsulation in the Novamente design. Yes, yes, I saw it in the design . . . . a miracle happens here. Which, granted, is better than not realizes that the area exists . . . . but still . . . . There are specific algorithms proposed, in the NM book, for doing map encapsulation. You may not believe they will work for the task, but still, it's not fair to use the label a miracle happens here to describe a description of specific algorithms applied to a specific data structure. I do not think the design has any huge gaps. But much further RD work is required
Re: [agi] What best evidence for fast AI?
That's a good simple, starting case. But how do you decide how much knowledge to disburse? How do you know what is irrelevant? How much do your answers differ between a small farmer in New Zealand, a rodeo rider in the West, a veterinarian is Pennsylvania, a child in Washington, a bio-mechanician studying gait? And horse is actually a *really* simple concept since it refers to a very specific type of physical object. Besides, are you really claiming that you'll be able to do this next year? Sorry, but that is just plain, unadulterated BS. If you can do that, you are light-years further along than . . . . Actually, this example is just not that hard. I think we may be able to do this during 2008, if funding for that particular NM application project holds up (it's currently confirmed only thru May-June) ben - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64228510-369311
RE: [agi] What best evidence for fast AI?
Ben, Thanks, I think Mark is raising some interesting issues. I may not agree with him on all of them but it is good to have your ideas tested by intelligent questioning. Ed -Original Message- From: Benjamin Goertzel [mailto:[EMAIL PROTECTED] Sent: Monday, November 12, 2007 11:37 AM To: agi@v2.listbox.com Subject: Re: [agi] What best evidence for fast AI? Ed -- Just a quick comment: Mark actually read a bunch of the proprietary, NDA-required Novamente documents and looked at some source code (3 years ago, so a lot of progress has happened since then). Richard didn't, so he doesn't have the same basis of knowledge to form detailed comments on NM, that Mark does. -- Ben On Nov 12, 2007 11:35 AM, Edward W. Porter [EMAIL PROTECTED] wrote: I'm sorry. I guess I did misunderstand you. If you have time I wish you could state the reasons why you find it lacking as efficiently as has Mark Waser. Ed Porter -Original Message- From: Richard Loosemore [mailto:[EMAIL PROTECTED] Sent: Monday, November 12, 2007 11:20 AM To: agi@v2.listbox.com Subject: Re: [agi] What best evidence for fast AI? Edward W. Porter wrote: Richard Loosemore wrote in a Sun 11/11/2007 11:09 PM post RICHARD You are right. I have only spent about 25 years working on this problem. Perhaps, no matter how bright I am, this is not enough to understand Novamente's promise. ED There a many people who have spent 25 years working on AI who have not spent the time to try to understand the multiple threads that make up the Novamente approach. From the one paper I read from you, as I remember it, your major approach to AI was based on a concept of complexity in which it was hard-for-humans-to-understand the relationship between the lower level of the system and the higher level functions you presumably want it to have. This is very different than the Novamente approach, which involves complexity, but not so much at an architectural level, but rather at the level of what will emerge in the self-organizing gen/comp network of patterns and behaviors that architecture is designed to grow, all under the constant watchful eye -- and selective weeding and watering -- of its goal and reward systems. As I understand it, the complexity in Novamente is much more like that in an economy in which semi-rational actors struggle to find and make a niche at which they can make a living, than the somewhat more anarchical complexity in the cellular automata Game Of Life. I am sorry, but this is a rather enormous misunderstanding of the claim I made. Too extensive for me to be able to deal with in a list post. So perhaps you are like most people who have spent a career in AI, in that the deep learning you have obtained has not spend enough time thinking about the pieces of Novamente-like approaches. But it is almost certain that that 25 years worth of knowledge would make it much easier for you to understand Novamente-like approach than all but a very small percent of this planet/s people, if you really wanted to. ED I am sure you are smart enough to understand its promise if you wanted to. Do you? RICHARD I did want to. I did. I do. ED Great. If you really do, I would start reading the papers at ___http://www.novamente.net/papers/_. Perhaps Ben could give you a better reading list than I. I don't know about you, Richard, but given my mental limitations, I often find I have to read some parts of paper 2 to 10 times to understand them. Usually much is unsaid in most papers, even the well written ones. You often have to spend time filling in the blanks and trying to imagine how what its describing would actually work. Much of my understanding of the Novamente approach not only comes from a broad range of reading and attending lectures in AI, micro-electronic, and brain science, but also a lot of thinking about what I have read and heard from other, and about what I have observed over decades of my own thought processes. There is a fundamental misunderstanding here, Ed. I read all of the Novamente papers a couple of years ago. My own thinking had already gone to that point and (in my opinion) well beyond it. You are implying that perhaps I do not understand it well enough. I understand it, understand a very wide range of issues that surround it, and also understand what i see as some serious limitations (some of which are encapsulated in my complexity paper). Thanks for your concern, but understanding the Novamente approach is not my problem. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/? http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/? http://v2.listbox.com/member/?; _ This list is sponsored by AGIRI
Re: [agi] What best evidence for fast AI?
On Nov 12, 2007 1:49 PM, Mark Waser [EMAIL PROTECTED] wrote: I'm more interested at this stage in analogies like -- btw seeking food and seeking understanding -- between getting an object out of a hole and getting an object out of a pocket, or a guarded room Why would one need to introduce advanced scientific concepts to an early-stage AGI? I don't get it... :-) A bit disingenuous there, Ben. Obviously you start with the simple and move on to the complex (though I suspect that the first analogy you cite is rather more complex than you might think) -- but to take too simplistic an approach that might not grow is just the narrow AI approach in other clothing. Well, I don't think we're doing the latter, obviously. It's not as though we are creating an AGI architecture that is overfitted to controlling simple organisms in virtual worlds. We've created a general AGI architecture and will then be applying it in this particular context. Hmmm I guess I didn't understand what you meant. What I thought you meant was, if a user asked I'm a small farmer in New Zealand. Tell me about horses then the system would be able to disburse its relevant knowledge about horses, filtering out the irrelevant stuff. What did you mean, exactly? That's a good simple, starting case. But how do you decide how much knowledge to disburse? How do you know what is irrelevant? How much do your answers differ between a small farmer in New Zealand, a rodeo rider in the West, a veterinarian is Pennsylvania, a child in Washington, a bio-mechanician studying gait? And horse is actually a *really* simple concept since it refers to a very specific type of physical object. Besides, are you really claiming that you'll be able to do this next year? Sorry, but that is just plain, unadulterated BS. If you can do that, you are light-years further along than . . . . Well, understanding the relevant context underlying a query is a fuzzy, not an absolute thing. There can be varying levels of capability at doing this. We have the basic mechanisms to enable this in NM, but they won't during 2008 perform this kind of contextualization as well as humans do. I didn't mean to be implying they would. There are specific algorithms proposed, in the NM book, for doing map encapsulation. You may not believe they will work for the task, but still, it's not fair to use the label a miracle happens here to describe a description of specific algorithms applied to a specific data structure. I guess that the jury will have to be out until you publicize the algorithms. What I've seen in the past are too small, too simple, and won't scale to what is likely to be necessary. I disagree, but this would get into a very in-depth technical conversation which isn't really apropos for this list. I think it has medium-sized gaps, not huge ones. I have not filled all these gaps because of lack of time -- implementing stuff needs to be balanced with finalizing design details of stuff that won't be implemented for a while anyway due to limited resources. :-) You have more than enough design experience to know that medium-size gaps can frequently turn huge once you turn your attention to them. Who are you snowing here? Certainly they can, but I've thought about these particular gaps a lot, and believe that's not going to happen here. But of course it **could** -- as I keep saying, completing the NM system does involve some RD, not pure engineering. -- Ben G - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64229523-f67219
Re: [agi] What best evidence for fast AI?
I don't know at what point you'll be blocked from answering by confidentiality concerns but I'll ask a few questions you hopefully can answer like: 1.. How is the information input and stored in your system (i.e. Is it more like simple formal assertions with a restricted syntax and/or language or like English language)? 2.. How constrained in the information content (and is the content even relevant)? 3.. To what degree does the system understand the information (i.e. how much can in manipulate it)? 4.. Who tags the information as relevant to particular users? 5.. How constrained are the tags? 6.. What is the output (is it just a regurgitation of appropriately tagged information pieces)? I have to assume that you're taking the easy way out on most of the questions (like formal assertions, restricted syntax, any language but the system does not understand or manipulate the language so content is irrelevant, users apply tags, fairly simply regurgitation) if you think 2008 is anywhere close to reasonable. - Original Message - From: Benjamin Goertzel To: agi@v2.listbox.com Sent: Monday, November 12, 2007 1:59 PM Subject: Re: [agi] What best evidence for fast AI? That's a good simple, starting case. But how do you decide how much knowledge to disburse? How do you know what is irrelevant? How much do your answers differ between a small farmer in New Zealand, a rodeo rider in the West, a veterinarian is Pennsylvania, a child in Washington, a bio-mechanician studying gait? And horse is actually a *really* simple concept since it refers to a very specific type of physical object. Besides, are you really claiming that you'll be able to do this next year? Sorry, but that is just plain, unadulterated BS. If you can do that, you are light-years further along than . . . . Actually, this example is just not that hard. I think we may be able to do this during 2008, if funding for that particular NM application project holds up (it's currently confirmed only thru May-June) ben -- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64259017-2fd868
Re: [agi] What best evidence for fast AI?
On Nov 12, 2007 2:51 PM, Mark Waser [EMAIL PROTECTED] wrote: I don't know at what point you'll be blocked from answering by confidentiality concerns I can't say much more than I will do in this email, due to customer confidentiality concerns but I'll ask a few questions you hopefully can answer like: 1. How is the information input and stored in your system (i.e. Is it more like simple formal assertions with a restricted syntax and/or language or like English language)? English input as well as other forms of input; NM Atom storage Obviously English language comprehension will not be complete; and proprietary (not Novamente's) UI devices will be used to work around this. 1. 2. How constrained in the information content (and is the content even relevant)? We'll work with a particular (relatively simple) text source for starters, with a view toward later generalization 1. 2. To what degree does the system understand the information (i.e. how much can in manipulate it)? That degree will increase as we bring more and more of PLN into the system. Initially, it'll just be simple PLN first-order term logic inference; then we'll extend it. 1. 2. Who tags the information as relevant to particular users? User feedback 1. 2. How constrained are the tags? They're English 1. 2. What is the output That's confidential, but it's very expressive and flexible | (is it just a regurgitation of appropriately tagged information pieces)? No -- Ben - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64260324-14ecdf
Re: [agi] What best evidence for fast AI?
On Nov 12, 2007 2:41 PM, Mark Waser [EMAIL PROTECTED] wrote: It is NOT clear that Novamente documentation is NOT enabling, or could not be made enabling, with, say, one man year of work. Strong argument could be made both ways. I believe that Ben would argue that Novamente documentation is NOT enabling even with one man-year of work. Ben? There is still way to much *research* work to be done. I'm not really familiar with this terminology, and don't have time to study it right now. But the standard for non-enablement is very arguably weaker than not requiring a miracle. It would be more like not requiring a leap of creativity that is outside the normal skill of talented PhDs trained in related fields. Yes. I believe that completion of NM does not require any leaps of creativity outside the normal skill of talented PhD's trained in related fields. Ask Ben how much actual work has been done on activation control in very large, very sparse atom spaces in Novamente. He'll tell you that it's a project for when he's further along. In this regard you are a bit out of date, Mark, due to your lack of recent contact w/ the NM project. In 2005 we did some testing of NM attention allocation mechanisms w/ millions of nodes and hundreds of millions of links, derived from NLP parsing and quantitative data mining. More recently I did some smaller-scale testing of similar (but better) mechanisms in a Ruby prototype, but this code is not yet ported into the main C++ codebase. This was all researchy stuff done with throwaway code just to see how the math worked on large AtomTables. But testing these mechanisms in isolation is not that informative -- they seem to work, but the real test will be seeing how they work in combination with large-scale inference and evolutionary learning, and we're not ready for that yet, due to incompleteness of the PLN and MOSES codebases relative to the respective designs. -- Ben - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64260364-265a64
Re: [agi] What best evidence for fast AI?
Hmm. Interesting. This e-mail (and the last) lead me to guess that you seem to have made some major, quantum leaps in NLP. Is that correct? You sure haven't been talking about it . . . . - Original Message - From: Benjamin Goertzel To: agi@v2.listbox.com Sent: Monday, November 12, 2007 2:57 PM Subject: Re: [agi] What best evidence for fast AI? On Nov 12, 2007 2:51 PM, Mark Waser [EMAIL PROTECTED] wrote: I don't know at what point you'll be blocked from answering by confidentiality concerns I can't say much more than I will do in this email, due to customer confidentiality concerns but I'll ask a few questions you hopefully can answer like: 1.. How is the information input and stored in your system (i.e. Is it more like simple formal assertions with a restricted syntax and/or language or like English language)? English input as well as other forms of input; NM Atom storage Obviously English language comprehension will not be complete; and proprietary (not Novamente's) UI devices will be used to work around this. 1.. 2.. How constrained in the information content (and is the content even relevant)? We'll work with a particular (relatively simple) text source for starters, with a view toward later generalization 1.. 2.. To what degree does the system understand the information (i.e. how much can in manipulate it)? That degree will increase as we bring more and more of PLN into the system. Initially, it'll just be simple PLN first-order term logic inference; then we'll extend it. 1.. 2.. Who tags the information as relevant to particular users? User feedback 1.. 2.. How constrained are the tags? They're English 1.. 2.. What is the output That's confidential, but it's very expressive and flexible | (is it just a regurgitation of appropriately tagged information pieces)? No -- Ben -- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64263051-2c4067
Re: [agi] What best evidence for fast AI?
Benjamin Goertzel wrote: Ed -- Just a quick comment: Mark actually read a bunch of the proprietary, NDA-required Novamente documents and looked at some source code (3 years ago, so a lot of progress has happened since then). Richard didn't, so he doesn't have the same basis of knowledge to form detailed comments on NM, that Mark does. This is true, but not important to my line of argument, since of course I believe that a problem exists (CSP), which we have discussed on a number of occasions, and your position is not that you have some proprietary, unknown-to-me solution to the problem, but rather that you do not really think there is a problem. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64272026-c0e7dd
Re: [agi] What best evidence for fast AI?
On Sat, Nov 10, 2007 at 10:19:44AM -0800, Jef Allbright wrote: as I was driving home I approached a truck off the side of the road, its driver pulling hard on a bar, tightening the straps securing the load. Without conscious thought I moved over in my lane to allow for the possibility that he might slip. That chain of inference, and its requisite knowledge base, leading to a simple human behavior, are not even on the radar horizon of current AI technology. ? I see a human, better give him wide berth. Certainly, the ability to detect and deal with pedestrians will be required before these things become street-legal. I can easily imagine that next-years grand challenge, or the one thereafter, will explicitly require ability to deal with cyclists, motorcyclists, pedestrians, children and dogs. Exactly how they'd test this, however, I don't know ... --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64316665-a9fb25
Re: [agi] What best evidence for fast AI?
On Nov 12, 2007 10:34 PM, Linas Vepstas [EMAIL PROTECTED] wrote: I can easily imagine that next-years grand challenge, or the one thereafter, will explicitly require ability to deal with cyclists, motorcyclists, pedestrians, children and dogs. Exactly how they'd test this, however, I don't know ... DARPA seems to be winding up the car challenges :-( (anyone knows anything to the contrary?) - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64332374-2a763e
Re: [agi] What best evidence for fast AI?
Linas Vepstas wrote: On Sat, Nov 10, 2007 at 10:19:44AM -0800, Jef Allbright wrote: as I was driving home I approached a truck off the side of the road, its driver pulling hard on a bar, tightening the straps securing the load. Without conscious thought I moved over in my lane to allow for the possibility that he might slip. That chain of inference, and its requisite knowledge base, leading to a simple human behavior, are not even on the radar horizon of current AI technology. ? I see a human, better give him wide berth. Certainly, the ability to detect and deal with pedestrians will be required before these things become street-legal. I can easily imagine that next-years grand challenge, or the one thereafter, will explicitly require ability to deal with cyclists, motorcyclists, pedestrians, children and dogs. Exactly how they'd test this, however, I don't know ... The problem (essentially the frame problem) is that it is no good to say Oh, we had better code for the situation of avoiding pedestrians, cyclists, children and dogs, it is that the system needs to be able to generally model the world in such a way that it can *anticipate*, by itself, a general situation that looks like developing into a problem. You never know what new situation might arise that might be a problem, and you cannot market a driverless car on the understanding that IF it starts killing people under particular circumstances, THEN someone will follow that by adding code to deal with that specific circumstance. The whole question then becomes: just how general are the mechanisms for understanding that a situation is a problem situation (like the one that Jef posed)? My understanding of the existing technology is that it is ridiculously far from being able to represent the world in such a general way that it could anticipate novel hazards without using up too many pedestrians. Absent that solution, I don't think these systems are going to be onthe market any time soon. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64343296-de95d4
Re: [agi] What best evidence for fast AI?
On 11/12/07, Linas Vepstas [EMAIL PROTECTED] wrote: I see a human, better give him wide berth. Certainly, the ability to detect and deal with pedestrians will be required before these things become street-legal. Well, I think we'll see robotic vehicles first play a significant role in war zones (including populated urban settings) with flashing lights and audible warning devices advising bystanders of their responsibility to avoid the risk. A difficulty (and this is only my limited, personal opinion) is that as the problems become more subtle, the corresponding requirements for extended inference increase exponentially. But I realize that what we're talking about here are really subtle problems, as in really quite small. I can easily imagine that next-years grand challenge, or the one thereafter, will explicitly require ability to deal with cyclists, motorcyclists, pedestrians, children and dogs. Exactly how they'd test this, however, I don't know ... Well it's clear from this and an earlier post of yours today that you (among relatively few others here) have a sound grasp of the big picture, and anything remaining is just minor detail. Makes me wonder why I tend to make everything so complicated. - Jef - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64347199-d76b50
Re: [agi] What best evidence for fast AI?
On 11/12/07, Lukasz Stafiniak [EMAIL PROTECTED] wrote: On Nov 12, 2007 10:34 PM, Linas Vepstas [EMAIL PROTECTED] wrote: I can easily imagine that next-years grand challenge, or the one thereafter, will explicitly require ability to deal with cyclists, motorcyclists, pedestrians, children and dogs. Exactly how they'd test this, however, I don't know ... DARPA seems to be winding up the car challenges :-( (anyone knows anything to the contrary?) There's no word of a further event, and no buzz, but plenty of similar question at the event. But if it's any consolation to you, Singapore has a grand challenge in the works involving robots able to enter buildings, operate doors, elevators, etc. and use weapons (only for defense, of course.) - Jef - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64349751-a59bd4
Re: [agi] What best evidence for fast AI?
I'm going to try to put some words into Richard's mouth here since I'm curious to see how close I am . . . . (while radically changing the words). I think that Richard is not arguing about the possibility of Novamente-type solutions as much as he is arguing about the predictability of *very* flexible Novamente-type solutions as they grow larger and more complex (and the difficulty in getting it to not instantaneously crash-and-burn). Indeed, I have heard a very faint shadow of Richard's concerns in your statements about the tuning problems that you had with BioMind. Novamente looks, at times, like the very first step in an inductive proof . . . . except that it is in a chaotic environment rather than the nice orderly number system. Pieces of the system clearly sail in calm, friendly waters but hooking them all up in a wild environment is another story entirely (again, look at your own BioMind stories). I've got many doubts because I don't think that you have a handle on the order -- the big (O) -- of many of the operations you are proposing (why I harp on scalability, modularity, etc.). Richard is going further and saying that the predictability of even some of your smaller/simpler operations is impossible (although, as he has pointed out, many of them could be constrained by attractors, etc. if you were so inclined to view/treat your design that way). Personally, I believe that intelligence is *not* complex -- despite the fact that it does (probably necessarily) rest on top of complex pieces -- because those pieces' interactions are constrained enough that intelligence is stable. I think that this could be built into a Novamente-type design *but* you have to be attempting to do so (and I think that I could convince Richard of that -- or else, I'd learn a lot by trying :-). Richard's main point is that he believes that the search space of viable parameters and operations for Novamente is small enough that you're not going to hit it by accident -- and Novamente's very flexibility is what compounds the problem. Remember, life exists on the boundary between order and chaos. Too much flexibility (unconstrained chaos) is as deadly as too much structure. I think that I see both sides of the issue and how Novamente could be altered/enhanced to make Richard happy (since it's almost universally flexible) -- but doing so would also impose many constraints that I think that you would be unwilling to live with since I'm not sure that you would see the point. I don't think that you're ever going to be able to change his view that the current direction of Novamente is -- pick one: a) a needle in an infinite haystack or b) too fragile to succeed -- particularly since I'm pretty sure that you couldn't convince me without making some serious additions to Novamente - Original Message - From: Benjamin Goertzel To: agi@v2.listbox.com Sent: Monday, November 12, 2007 3:49 PM Subject: Re: [agi] What best evidence for fast AI? To be honest, Richard, I do wonder whether a sufficiently in-depth conversation about AGI between us would result in you changing your views about the CSP problem in a way that would accept the possibility of Novamente-type solutions. But, this conversation as I'm envisioning it would take dozens of hours, and would require you to first spend 100+ hours studying detailed NM materials, so this seems unlikely to happen in the near future. -- Ben On Nov 12, 2007 3:32 PM, Richard Loosemore [EMAIL PROTECTED] wrote: Benjamin Goertzel wrote: Ed -- Just a quick comment: Mark actually read a bunch of the proprietary, NDA-required Novamente documents and looked at some source code (3 years ago, so a lot of progress has happened since then). Richard didn't, so he doesn't have the same basis of knowledge to form detailed comments on NM, that Mark does. This is true, but not important to my line of argument, since of course I believe that a problem exists (CSP), which we have discussed on a number of occasions, and your position is not that you have some proprietary, unknown-to-me solution to the problem, but rather that you do not really think there is a problem. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; -- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64351025-209479
Re: [agi] What best evidence for fast AI?
On Sun, Nov 11, 2007 at 02:16:06PM -0500, Edward W. Porter wrote: Its way out, but not crazy. If humanity or some mechanical legacy of us ever comes out the other end of the first century after superhuman intelligence arrives, it or they will be ready to start playing in the Galactic big leagues. Or, if Nick Bostrom is right about his simulation argument, then perhaps instead our simulators will reveal themselves to us. So far, I find Bostrom's work as one of the more reasonable solutions to the Fermi paradox ('where are they?'). --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64351589-c930a4
Re: [agi] What best evidence for fast AI?
On Mon, Nov 12, 2007 at 04:56:00PM -0500, Richard Loosemore wrote: Linas Vepstas wrote: I can easily imagine that next-years grand challenge, or the one thereafter, will explicitly require ability to deal with cyclists, motorcyclists, pedestrians, children and dogs. Exactly how they'd test this, however, I don't know ... The problem (essentially the frame problem) is that it is no good to say Oh, we had better code for the situation of avoiding pedestrians, cyclists, children and dogs, it is that the system needs to be able to generally model the world in such a way that it can *anticipate*, by itself, a general situation that looks like developing into a problem. Yes, but there is a standard solution for the frame problem that has been in use for several decades now. Its those signs posted on highway entrance ramps that state Minimum speed 45 mph. Bicycle and pedestrian access prohibited. I hate to explain flip answers, but sigh, I guess I need to sometimes. I'm saying that the solution to the frame problem can sometimes be to not solve it. My cognition and perception abilites are not so great as to be able to avoid being hit by a meteor as I drive down the highway: in other words, my brain fails to solve that particular frame problem as well. It is also somewhat unprepared for mexican trucks with bad brakes and bald tires, and so the standard solution is to make these illegal. Human beings, when college educated, can sometimes *anticipate* a general situation that looks like developing into a problem, but not always, and usually not at highway speeds. --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64359579-cb8713
Re: [agi] What best evidence for fast AI?
You seem to be thinking about Webmind, an AI company I was involved in during the late 1990's; as opposed to Biomind Yes, sorry, I'm laboring under a horrible cold and my brain is not all here. The big-O order is almost always irrelevant. Most algorithms useful for cognition are exponential-time worst-case complexity. What matters is average-case complexity over the probability distribution of problem instances actually observed in the real world. And yeah, this is very hard to estimate mathematically. Well . . . . big-O order certainly does matter for things like lookups and activation where we're not talking about heuristic shortcuts and average complexity. But I would certainly accept your correction for other operations like finding modularity and analogies -- except we don't have good heuristic shortcuts, etc. for them -- yet. Saying a system is universally capable doesn't mean hardly anything, and isn't really worth saying. Nope. Saying it usually forestalls a lot of silly objections. That's really worthwhile.:-) I believe Richard's complaints are of a quite different character than yours. And I might be projecting . . . . :-)which is why I figured I'd run this out there and see how he reacted.:-) - Original Message - From: Benjamin Goertzel To: agi@v2.listbox.com Sent: Monday, November 12, 2007 5:14 PM Subject: Re: [agi] What best evidence for fast AI? On Nov 12, 2007 5:02 PM, Mark Waser [EMAIL PROTECTED] wrote: I'm going to try to put some words into Richard's mouth here since I'm curious to see how close I am . . . . (while radically changing the words). I think that Richard is not arguing about the possibility of Novamente-type solutions as much as he is arguing about the predictability of *very* flexible Novamente-type solutions as they grow larger and more complex (and the difficulty in getting it to not instantaneously crash-and-burn). Indeed, I have heard a very faint shadow of Richard's concerns in your statements about the tuning problems that you had with BioMind. You seem to be thinking about Webmind, an AI company I was involved in during the late 1990's; as opposed to Biomind, a bioinformatics company in which I am currently involved, and which is doing pretty well. The Webmind AI Engine was an order of magnitude more complex than the Novamente Cognition Engine; and this is intentional. Many aspects of the NM design were specifically originated to avoid problems that we found with the Webmind system. I've got many doubts because I don't think that you have a handle on the order -- the big (O) -- of many of the operations you are proposing (why I harp on scalability, modularity, etc.). The big-O order is almost always irrelevant. Most algorithms useful for cognition are exponential-time worst-case complexity. What matters is average-case complexity over the probability distribution of problem instances actually observed in the real world. And yeah, this is very hard to estimate mathematically. Richard is going further and saying that the predictability of even some of your smaller/simpler operations is impossible (although, as he has pointed out, many of them could be constrained by attractors, etc. if you were so inclined to view/treat your design that way). Oh, I thought **I** was the one who pointed that out. Personally, I believe that intelligence is *not* complex -- despite the fact that it does (probably necessarily) rest on top of complex pieces -- because those pieces' interactions are constrained enough that intelligence is stable. I think that this could be built into a Novamente-type design *but* you have to be attempting to do so (and I think that I could convince Richard of that -- or else, I'd learn a lot by trying :-). That is part of the plan, but we have a bunch of work of implementing/tuning components first. Richard's main point is that he believes that the search space of viable parameters and operations for Novamente is small enough that you're not going to hit it by accident -- and Novamente's very flexibility is what compounds the problem. The Webmind system had this problem. Novamente is carefully designed not to. Of course, I can't prove that it won't, though. Remember, life exists on the boundary between order and chaos. Too much flexibility (unconstrained chaos) is as deadly as too much structure. I think that I see both sides of the issue and how Novamente could be altered/enhanced to make Richard happy (since it's almost universally flexible) -- Novamente is universally capable but so are a lot of way simpler, pragmatically useless system. Saying a system is universally capable doesn't mean hardly anything, and isn't really worth saying. The question as you know is what can a system do given a pragmatic amount
Re: [agi] What best evidence for fast AI?
I am heavily focussed on my own design at the moment, but when you talk about the need for 100+ hours of studying detailed NM materials, are you talking about publicly available documents, or proprietary information? Proprietary info, much of which may be made public next year, though... - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64372765-448103
Re: [agi] What best evidence for fast AI?
Benjamin Goertzel wrote: To be honest, Richard, I do wonder whether a sufficiently in-depth conversation about AGI between us would result in you changing your views about the CSP problem in a way that would accept the possibility of Novamente-type solutions. But, this conversation as I'm envisioning it would take dozens of hours, and would require you to first spend 100+ hours studying detailed NM materials, so this seems unlikely to happen in the near future. Well, I am not by any means hostile to the idea that Novamente could be built in such a way as to solve the CSP. It is all a question of methodology and flexibility, which I don't *think* is there, but I could be wrong. I am heavily focussed on my own design at the moment, but when you talk about the need for 100+ hours of studying detailed NM materials, are you talking about publicly available documents, or proprietary information? Richard Loosemore -- Ben On Nov 12, 2007 3:32 PM, Richard Loosemore [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote: Benjamin Goertzel wrote: Ed -- Just a quick comment: Mark actually read a bunch of the proprietary, NDA-required Novamente documents and looked at some source code (3 years ago, so a lot of progress has happened since then). Richard didn't, so he doesn't have the same basis of knowledge to form detailed comments on NM, that Mark does. This is true, but not important to my line of argument, since of course I believe that a problem exists (CSP), which we have discussed on a number of occasions, and your position is not that you have some proprietary, unknown-to-me solution to the problem, but rather that you do not really think there is a problem. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; http://v2.listbox.com/member/?; This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64373890-c05dbe
Re: [agi] What best evidence for fast AI?
On Mon, Nov 12, 2007 at 01:49:52PM -0500, Mark Waser wrote: What I thought you meant was, if a user asked I'm a small farmer in New Zealand. Tell me about horses then the system would be able to disburse its relevant knowledge about horses, filtering out the irrelevant stuff. What did you mean, exactly? That's a good simple, starting case. But how do you decide how much knowledge to disburse? How do you know what is irrelevant? How much do your answers differ between a small farmer in New Zealand, a rodeo rider in the West, a veterinarian is Pennsylvania, a child in Washington, a bio-mechanician studying gait? And horse is actually a *really* simple concept since it refers to a very specific type of physical object. Besides, are you really claiming that you'll be able to do this next year? Sorry, but that is just plain, unadulterated BS. If you can do that, you are light-years further along than . . . . Eh? I can demo a system to you today, that does a very lame version of this. And it's probably only the umpteenth system to do this, and it does it in only a few thousand lines of code (not counting modules pulled off the net). Its a bot on #opencyc on freenode.net (seems to be crashed at the moment) When you ask it about Abraham Lincoln, it will respond with a grade-school like essay that Abe is a person and a male person and a historical person and is famous. All it knows is from the opencyc db. It will happily include irrelevant facts like Abe is a person and a male person, but it has some ability to prune these; when you ask again, it'll refuse to answer, with an I already told you response. Its not AI, but it does demonstrate those things you are calling BS. As to talking about horses, even I am not capable of maintaining a conversation with a rodeo rider, and I live in Texas. I once talked to a professional blacksmith; turns out they are required by law to have a degree in veternary medicine; bet you didn't know that. If and when you find a human who is capable of having conversations about horses with small farmers, rodeo riders, vets, children and biomechanicians, I'll bet that they won't have a clue about galaxy formation or enzyme reactions. Don't set the bar above human capabilites. --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64386951-c91d87
Re: [agi] What best evidence for fast AI?
On Monday 12 November 2007 15:56, Richard Loosemore wrote: You never know what new situation might arise that might be a problem, and you cannot market a driverless car on the understanding that IF it starts killing people under particular circumstances, THEN someone will follow that by adding code to deal with that specific circumstance. It seems that this was the way that the brain was progressively 'improved' via evolution. However, we want to compress a few billion years of evolutionary selective pressure into the next 10 or 100 years instead. Have there been any proposed strategies that try to take an evolutionary approach on the magnitude that was needed for human brain evolution? - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64395703-74a5cb
Re: [agi] What best evidence for fast AI?
On Mon, Nov 12, 2007 at 06:56:51PM -0500, Mark Waser wrote: It will happily include irrelevant facts Which immediately makes it *not* relevant to my point. Please read my e-mails more carefully before you hop on with ignorant flames. I read your emails, and, mixed in with some insightful and highly relevent commentary, there are also many flames. Repeatedly so. Relevence is not an easy problem, nor is it obviously a hard one. To provide relevent answers, one must have a model of who is asking. So, in building a computer chat system, one must first deduce things about the speaker. This is something I've been trying to do. Again, with my toy system, I've gotten so far as to be able to let the speaker proclaim that this is boring, and have the system remember, so that, for future conversations, the boring assertions are not revisited. Now, boring is a tricky thing: a horse is genus equus may be boring for a child, and yet interesting to young adults. So the problem of relevent answers to questions is more about creating a model of the person one is conversing with, than it is about NLP processing, representation of knowledge, etc. Conversations are contextual; modelling that context is what is interesting to me. The result of hooking up a reasoning system, a knowledgebase like opencyc or sumo, an nlp parser, and a homebrew contextualizer is not agi. It's little more than a son-et-lumiere show. But it already does the things that you are claiming to be unadulterated BS. And regarding If and when you find a human who is capable of having conversations about horses with small farmers, rodeo riders, vets, children and biomechanicians, I'll bet that they won't have a clue about galaxy formation or enzyme reactions. Don't set the bar above human capabilites. Go meet your average librarian. They won't know the information off the top of their heads (yet), but they'll certainly be able to get it to you -- Go meet google. Or wikipedia. Cheeses. and the average librarian fifteen years from now *will* be able to. When the average librarian is able to answer veterinary questions to the satisfaction of a licensing board conducting an oral examination, then we will be living in the era of agi, won't we? --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64410937-e020ba
Re: [agi] What best evidence for fast AI?
:-) I don't think I've ever known you to spout intentionally BS . . . . A well-architected statistical-NLP-based information-retrieval system would require an identification (probably an exemplar) of the cluster(s) that matched each of the portfolios and would return a mixed conglomerate of data rather than any sort of coherent explanation (other than the explanations present in the data cluster). The WASNLPBIRS certainly wouldn't be able to condense the data to a nicely readable format or perform any other real operations on the information. What I meant by *really* sophisticated should have been indicated by the difficult end of my six point list -- which is fundamentally equivalent (in my opinion) to a full-up AGI since it basically requires full understanding of English and a WASNLPBIRS feeding it. The problem with the WASNLPBIRS and what Linas suggested is that they look *really* cool at first -- and then you realize how little they actually do. The real problem with your claim of if a user asked I'm a small farmer in New Zealand. Tell me about horses then the system would be able to disburse its relevant knowledge about horses, filtering out the irrelevant stuff is the last five words. How do you intend to do *that*. (And notice that what I kicked Linas for was precisely his It will happily include irrelevant facts. I've had to deal with users who have bought large, expensive conceptual clustering systems who were *VERY* unhappy once they realized what they had actually purchased. I would be *real* careful if I were you about what you're promising because there are already a good number of companies that, a decade ago, had already perfected the best that that approach could offer -- and then died on the rope of user dissatisfaction. Mark - Original Message - From: Benjamin Goertzel To: agi@v2.listbox.com Sent: Monday, November 12, 2007 7:10 PM Subject: Re: [agi] What best evidence for fast AI? On Nov 12, 2007 6:56 PM, Mark Waser [EMAIL PROTECTED] wrote: It will happily include irrelevant facts Which immediately makes it *not* relevant to my point. Please read my e-mails more carefully before you hop on with ignorant flames. The latter part of your e-mail clearly makes my point -- anyone claiming to be able to do a sophisticated version of this in the next year is spouting plain, unadulterated BS. Mark, I really wasn't spouting BS. I imagine what you are conceiving when you use the label of sophisticated is more sophisticated than what I am hoping to launch within the next year. Being sophisticated is not a precise criterion. Your example of giving information about horses in a contextual way ** How do you know what is irrelevant? How much do your answers differ between a small farmer in New Zealand, a rodeo rider in the West, a veterinarian is Pennsylvania, a child in Washington, a bio-mechanician studying gait? ** is in my judgment not beyond what a well-architected statistical-NLP-based information-retrieval system could deliver. I don't think you even need a Novamente system to do this.So is this all you mean by sophisticated? I don't really understand what you intend... seriously... -- Ben -- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64413835-ad7189
Re: [agi] What best evidence for fast AI?
There is a big difference between being able to fake something for a brief period of time and being able to do it correctly. All of your phrasing clearly indicates that *you* believe that your systems can only fake it for a brief period of time, not do it correctly. Why are you belaboring the point? I don't get it since your own points seem to deny your own argument. And even if you can do it for small, toy conversations where you recognize the exact same assertions -- that is nowhere close to what you're going to need in the real world. When the average librarian is able to answer veterinary questions to the satisfaction of a licensing board conducting an oral examination, then we will be living in the era of agi, won't we? Depends upon your definition of AGI. That could be just a really kick-ass decision support system -- and I would actually bet a pretty fair chunk of money that 15 years *is* entirely within reason for the scenario you suggest. - Original Message - From: Linas Vepstas [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Monday, November 12, 2007 7:28 PM Subject: Re: [agi] What best evidence for fast AI? On Mon, Nov 12, 2007 at 06:56:51PM -0500, Mark Waser wrote: It will happily include irrelevant facts Which immediately makes it *not* relevant to my point. Please read my e-mails more carefully before you hop on with ignorant flames. I read your emails, and, mixed in with some insightful and highly relevent commentary, there are also many flames. Repeatedly so. Relevence is not an easy problem, nor is it obviously a hard one. To provide relevent answers, one must have a model of who is asking. So, in building a computer chat system, one must first deduce things about the speaker. This is something I've been trying to do. Again, with my toy system, I've gotten so far as to be able to let the speaker proclaim that this is boring, and have the system remember, so that, for future conversations, the boring assertions are not revisited. Now, boring is a tricky thing: a horse is genus equus may be boring for a child, and yet interesting to young adults. So the problem of relevent answers to questions is more about creating a model of the person one is conversing with, than it is about NLP processing, representation of knowledge, etc. Conversations are contextual; modelling that context is what is interesting to me. The result of hooking up a reasoning system, a knowledgebase like opencyc or sumo, an nlp parser, and a homebrew contextualizer is not agi. It's little more than a son-et-lumiere show. But it already does the things that you are claiming to be unadulterated BS. And regarding If and when you find a human who is capable of having conversations about horses with small farmers, rodeo riders, vets, children and biomechanicians, I'll bet that they won't have a clue about galaxy formation or enzyme reactions. Don't set the bar above human capabilites. Go meet your average librarian. They won't know the information off the top of their heads (yet), but they'll certainly be able to get it to you -- Go meet google. Or wikipedia. Cheeses. and the average librarian fifteen years from now *will* be able to. When the average librarian is able to answer veterinary questions to the satisfaction of a licensing board conducting an oral examination, then we will be living in the era of agi, won't we? --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64415771-2a51bf
Re: [agi] What best evidence for fast AI?
On Mon, Nov 12, 2007 at 06:22:37PM -0600, Bryan Bishop wrote: On Monday 12 November 2007 17:31, Linas Vepstas wrote: If and when you find a human who is capable of having conversations about horses with small farmers, rodeo riders, vets, children and biomechanicians, I'll bet that they won't have a clue about galaxy formation or enzyme reactions. Don't set the bar above human capabilites. Are these things supposed to be rare discussion topics? I think this just serves to illustrate the wide-ranging shades of normal that some of us see in the daily human population. This stuff is hard and we seem to restrict so much to one or two variables. Conversation is hard. You can talk to almost anyone about the weather, but you won't be able to talk to a rodeo rider about horses the way that other riders do. You can read a book about how to be a good conversationalist, apply the basic tricks it teaches you, with great success, and still remain ignorant and shallow. --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64418119-21691d
Re: [agi] What best evidence for fast AI?
On Mon, Nov 12, 2007 at 07:46:15PM -0500, Mark Waser wrote: There is a big difference between being able to fake something for a brief period of time and being able to do it correctly. All of your phrasing clearly indicates that *you* believe that your systems can only fake it for a brief period of time, not do it correctly. Why are you belaboring the point? I don't get it since your own points seem to deny your own argument. I don't think BenG claimed to be able to build an AGI in 6 months, but rather something that can fake it for a breif period of time. I was rising to the defense of that. When the average librarian is able to answer veterinary questions to the satisfaction of a licensing board conducting an oral examination, then we will be living in the era of agi, won't we? Depends upon your definition of AGI. That could be just a really kick-ass decision support system -- and I would actually bet a pretty fair chunk of money that 15 years *is* entirely within reason for the scenario you suggest. Actually, I agree with that. Or, to paraphrase, I think that NLP-speaking know-it-all librarians are reasonable in 15 years, as they seem to be just shiny and polished versions of things we have today. So perhaps the AGI question is, what is the difference between a know-it-all mechano-librarian, and a sentient being? --linas - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64420786-7a64ef
Re: [agi] What best evidence for fast AI?
Mark Waser wrote: Yes, sorry, I'm laboring under a horrible cold and my brain is not all here. Same here: I'm recovering from it now, but it was a real doozy. (Is that how you spell doozy?) Anyhow, this is all just to say that your detailed post and questions was very thought provoking, but it will have to wait until tomorrow for an answer Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64422194-d5c7a6
Re: [agi] What best evidence for fast AI?
Bryan Bishop wrote: On Monday 12 November 2007 15:56, Richard Loosemore wrote: You never know what new situation might arise that might be a problem, and you cannot market a driverless car on the understanding that IF it starts killing people under particular circumstances, THEN someone will follow that by adding code to deal with that specific circumstance. It seems that this was the way that the brain was progressively 'improved' via evolution. However, we want to compress a few billion years of evolutionary selective pressure into the next 10 or 100 years instead. Have there been any proposed strategies that try to take an evolutionary approach on the magnitude that was needed for human brain evolution? Yikes, no: my strategy is to piggyback on all that work, not to try to duplicate it. Even the Genetic Algorithm people don't (I think) dream of evolution on that scale. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64422731-1d8dab
Re: [agi] What best evidence for fast AI?
On Monday 12 November 2007 19:31, Richard Loosemore wrote: Yikes, no: my strategy is to piggyback on all that work, not to try to duplicate it. Even the Genetic Algorithm people don't (I think) dream of evolution on that scale. Yudkowsky recently wrote an email on preservation of the absurdity of the future. The method that I have proposed requires this massive international effort and maybe can only be started when we hit a few more billion births. It is not entirely absurd, however, since we would start the project with investigation methods known today and slowly improve until we have millions of people researching the millions of varied pathways in the brain. From what I have read of Novamente today, Goertzel might be hoping that the circuits in the brain are ultimately simple, or that some similar model that has simpler components building up to some greater actor-exchange medium, effectively mimics the brain to some degree. - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64423497-d1a153
Re: [agi] What best evidence for fast AI?
Edward W. Porter wrote: I'm sorry. I guess I did misunderstand you. If you have time I wish you could state the reasons why you find it lacking as efficiently as has Mark Waser. Ed Porter I'll do my best when I respond to Mark's questions/commentary tomorrow. Briefly, though, the complex systems paper I wrote really was my statement of the main problem (though seeing *how* it applies is, I admit, a rather big exercise for the reader). I suspect that because of the unusual nature of my claims about the complex systems problem, it will probably need a book-length exposition to make it clear. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64424505-67912a
Re: [agi] What best evidence for fast AI?
Bryan Bishop wrote: On Monday 12 November 2007 19:31, Richard Loosemore wrote: Yikes, no: my strategy is to piggyback on all that work, not to try to duplicate it. Even the Genetic Algorithm people don't (I think) dream of evolution on that scale. Yudkowsky recently wrote an email on preservation of the absurdity of the future. The method that I have proposed requires this massive international effort and maybe can only be started when we hit a few more billion births. It is not entirely absurd, however, since we would start the project with investigation methods known today and slowly improve until we have millions of people researching the millions of varied pathways in the brain. From what I have read of Novamente today, Goertzel might be hoping that the circuits in the brain are ultimately simple, or that some similar model that has simpler components building up to some greater actor-exchange medium, effectively mimics the brain to some degree. Yudkowsky's ramblings don't cut much ice with me. Ben is not so much interested in whether the circuits (mechanisms) in the brain are simple or not, since he belongs to the school that believes that AGI does not need to be done exactly the way the human mind does it. I, on the other hand, believe that we must stick fairly closely to an emulation of the *cognitive* level (not neural, but much higher up). Even with everyone on the planet running evolutionary simulations, I do not believe we could reinvent an intelligent system by brute force. Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64425716-c7bb52
Re: [agi] What best evidence for fast AI?
On Monday 12 November 2007 19:48, Richard Loosemore wrote: Even with everyone on the planet running evolutionary simulations, I do not believe we could reinvent an intelligent system by brute force. Of your message, this part is the most peculiar. Brute force is all that we have. - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64427651-dc0d91
Re: [agi] What best evidence for fast AI?
On Nov 12, 2007 8:44 PM, Mark Waser [EMAIL PROTECTED] wrote: I don't think BenG claimed to be able to build an AGI in 6 months, but rather something that can fake it for a breif period of time. I was rising to the defense of that. No. Ben is honest in his claims and he said that this was for a paying client. It isn't going to be a deliberate fake it for a brief period of time. He'll definitely deliver something cool -- I was much more objecting to some possibly dangerous, over-enthusiastic phrasing. Yes, for this NLP-related contract I mentioned, we are going for something cool in a limited domain, but built in a way allowing generalizability according to the NM architecture... My hope is that we won't get too bogged down in NLP particularities (as we've already got lots of code for handling this), and after 6 months or so we'll be able to spend most of our time on the project dealing with interesting PLN inference stuff. Also, the NLP code we make on this project will likely be integrable w/ our virtual-animal code, thus allowing us to create virtual embodied agents w/ linguistic ability as I've discussed before. -- Ben - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64428269-6457d0
Re: [agi] What best evidence for fast AI?
Bryan Bishop wrote: On Monday 12 November 2007 19:48, Richard Loosemore wrote: Even with everyone on the planet running evolutionary simulations, I do not believe we could reinvent an intelligent system by brute force. Of your message, this part is the most peculiar. Brute force is all that we have. We might be talking at cross purposes... I didn't intend to suggest that there was a brute-force and a non brute force way to duplicate evolution, with the brute force method being infeasible I was just trying to say that it would be such a gigantic project that I do not think it feasible. That's a bit of a judgment call, I guess, but since I think there are much more viable alternatives, I don't feel pressed to get a more accurate handle on just how difficult it would be. If anyone were to throw that quantity of resources at the AGI problem (recruiting all of the planet), heck, I could get it done in about 3 years. ;-) Richard Loosemore - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64448535-d88cd9
Re: [agi] What best evidence for fast AI?
On Monday 12 November 2007 22:16, Richard Loosemore wrote: If anyone were to throw that quantity of resources at the AGI problem (recruiting all of the planet), heck, I could get it done in about 3 years. ;-) I have done some research on this topic in the last hour and have found that a Connectome Project is in fact in the very early stages out there on the internet: http://iic.harvard.edu/projects/connectome.html http://acenetica.blogspot.com/2005/11/human-connectome.html http://acenetica.blogspot.com/2005/10/mission-to-build-simulated-brain.html http://www.indiana.edu/~cortex/connectome_plos.pdf - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64449857-7dd95a
RE: [agi] What best evidence for fast AI?
Ben said -- the possibility of dramatic, rapid, shocking success in robotics is LOWER than in cognition That's why I tell people the value of manual labor will not be impacted as soon by the AGI revolution as the value of mind labor. Ed Porter -Original Message- From: Benjamin Goertzel [mailto:[EMAIL PROTECTED] Sent: Saturday, November 10, 2007 5:29 PM To: agi@v2.listbox.com Subject: Re: [agi] What best evidence for fast AI? I'm impressed with the certainty of some of the views expressed here, nothing like I get talking to people actually building robots. - Jef Robotics involves a lot of difficulties regarding sensor and actuator mechanics and data-processing. Whether these need to be solved to create AGI is a matter of much contention. Some, like Rodney Brooks, think so. Others, like me, doubt it -- though I think embodiment does have a lot to offer an AGI system, hence my current focus on virtual embodiment... Still, in spite of the hurdles, the solvability of the problems facing humanoid robotics w/in the next few decades seems pretty clear to me --- if sufficient resources are devoted to the problem (and it's not clear they will be). I think that, compared to fundamental progress in AGI cognition, -- our certitude in dramatic robotics progress can be greater, under assumptions of adequate funding -- the possibility of dramatic, rapid, shocking success in robotics is LOWER than in cognition -- Ben G _ This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/? http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63975170-cc0347
Re: [agi] What best evidence for fast AI?
On 11/11/07, Edward W. Porter [EMAIL PROTECTED] wrote: Ben said -- the possibility of dramatic, rapid, shocking success in robotics is LOWER than in cognition That's why I tell people the value of manual labor will not be impacted as soon by the AGI revolution as the value of mind labor. Both valid points -- emphasizing possibility leading to dramatic, shocking success -- but this does not invalidate the (in my opinion) greater near-term *probability* of accelerating development and practical deployment of robotics and its broad impact on society. Robotics (like all physical technologies) will hit a ceiling defined by intelligence. Machine intelligence surpassing human capabilities in general will be far more dramatic, rapid, and shocking than any previous technology. But we do not yet have a complete, verifiable theory, let alone a practical design. - Jef - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63984519-51ebc9
Re: [agi] What best evidence for fast AI?
But we do not yet have a complete, verifiable theory, let alone a practical design. - Jef To be more accurate, we don't have a practical design that is commonly accepted in the AGI research community. I believe that I *do* have a practical design for AGI and I am working hard toward getting it implemented. This practical design is based on a theory that is fairly complete, but not easily verifiable using current technology. The verification, it seems, will come via actually getting the AGI built! -- Ben G - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63987650-f9a81b
Re: [agi] What best evidence for fast AI?
Richard, Even Ben Goertzel, in a recent comment, said something to the effect that the only good reason to believe that his model is going to function as advertised is that *when* it is working we will be able to see that it really does work: The above paragraph is a distortion of what I said, and misrepresents my own thoughts and beliefs. I think that, after the Novamente design and the ideas underlying it are carefully studied by a suitably trained individiual, the hypothesis that it will lead to a human-level AI comes to seem plausible. But, there is no solid proof, it's in part a matter of educated intuition. The following quote which you gave is accurate: Ben Goertzel wrote: This practical design is based on a theory that is fairly complete, but not easily verifiable using current technology. The verification, it seems, will come via actually getting the AGI built! This is a million miles short of a declaration that there are no hard problems left in AI. Whether there are hard problems left in AI, conditional on the assumption that the Novamente design is workable, comes down to a question of semantic interpretation. In the completion of the detailed-design and implementation of the Novamente system, there are around a half-dozen research problems on the PhD thesis level to be solved. This means there is some hard thinking left, yet if the Novamente design is correct, it pertains some well-defined and well-delimited technical questions, which seem very likely to be solvable. As an example, there is the task of generalizing the MOSES algorithm (see metacog.org) to handle general programmatic constructs at the nodes of its internal program trees. Of course this is a hard problem, yet it's a well-defined computer science problem which (after a lot of things) doesn't seem likely to be hiding any deep gotchas. But this is research and development -- not pure development -- so one never knows for sure... -- Ben - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64055433-fe7f04
Re: [agi] What best evidence for fast AI?
Benjamin Goertzel wrote: Richard, Even Ben Goertzel, in a recent comment, said something to the effect that the only good reason to believe that his model is going to function as advertised is that *when* it is working we will be able to see that it really does work: The above paragraph is a distortion of what I said, and misrepresents my own thoughts and beliefs. When pressed, you always resort to a phrase equivalent to the one you give below: I think that, after the Novamente design and the ideas underlying it are carefully studied by a suitably trained individiual, the hypothesis that it will lead to a human-level AI comes to seem plausible When you look carefully at this phrasing, its core is a statement that the best reason to believe that it will work is the *intuition* of someone who studies the design ... and you state that you believe that anyone who is suitably trained, who studies it, will have the same intuition that you do. This is all well and good, but it contains no metric, no new analysis of the outstanding problems that we can all scrutinize and assess. I would consider an appeal to the intuition of suitably trained individuals to be very much less than a good reason to believe that the model is going to function as advertised. Thus: if someone wanted volunteers to fly in their brand-new aircraft design, but all they could do to reassure people that it was going to work were the intuitions of suitably trained individuals, then most rational people would refuse to fly - they would want more than intuitions. In this light, my summary would not be a distortion of your position at all, but only a statement about whether an appeal to intuition counts as a good reason to believe. And, of course, there are some suitably trained individuals who do not share your intuitions, even given the limited access they have to your detailed design. I respect your optimism, and applaud your single-minded commitment to the project: if it is going to work, that is the way to get it done. I certainly wish you luck with it. Richard Loosemore I think that, after the Novamente design and the ideas underlying it are carefully studied by a suitably trained individiual, the hypothesis that it will lead to a human-level AI comes to seem plausible. But, there is no solid proof, it's in part a matter of educated intuition. The following quote which you gave is accurate: Ben Goertzel wrote: This practical design is based on a theory that is fairly complete, but not easily verifiable using current technology. The verification, it seems, will come via actually getting the AGI built! This is a million miles short of a declaration that there are no hard problems left in AI. Whether there are hard problems left in AI, conditional on the assumption that the Novamente design is workable, comes down to a question of semantic interpretation. In the completion of the detailed-design and implementation of the Novamente system, there are around a half-dozen research problems on the PhD thesis level to be solved. This means there is some hard thinking left, yet if the Novamente design is correct, it pertains some well-defined and well-delimited technical questions, which seem very likely to be solvable. As an example, there is the task of generalizing the MOSES algorithm (see metacog.org http://metacog.org) to handle general programmatic constructs at the nodes of its internal program trees. Of course this is a hard problem, yet it's a well-defined computer science problem which (after a lot of things) doesn't seem likely to be hiding any deep gotchas. But this is research and development -- not pure development -- so one never knows for sure... -- Ben This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64076724-00fae4
Re: [agi] What best evidence for fast AI?
Edward W. Porter wrote: Richard, Geortzel claims his planning indicates it is rougly 6 years x 15 excellent, hard-working programmers, or 90 man years to getting his architecture up an running. I assume that will involve a lot of “hard” mental work. By “hard problem” I mean a problem for which we don’t have what seems -- within the Novemente model -- to be a way for handling it at, at least, a roughly human-level. We won’t have proof that the problem is not hard until we actually get the part of the system that deals with that problem up and running successfully. Until then, you have every right to be skeptical. But you also have the right, should you so choose, to open your mind up to the tremendous potential of the Novamente approach. RICHARD What would be the solution of the grounding problem? ED Not hard. As one linguist said “Words are defined by the company they keep”. Kinda like I am guessing Google sets work, but at more different levels in the gen/comp pattern hierarchy and with more cross inferencing between different google-set seeds. The same goes not only for words, but for almost all concepts and sub-concepts. Grounding is made out of a life-time of experience recording such associations and the dynamic reactivation of those associations both in the subconscious and conscious in response to current activations. RICHARD What would be the solution of the problem of autonomous, unsupervised learning of concepts? ED Not hard! Read Novamente (or for a starter my prior summaries of it). That’s one of its main focus. RICHARD Can you find proofs that inference control engines will not show divergent behavior under heavy load (i.e. will they degrade gracefully when forced to provide answers in real time)? ED Not totally clear. Brain level hardware will really help here, but what is six orders of magnitude to the potential of combinatorial explosion in dynamic activations of something as large and high-dimensional as world knowledge?. This issue falls under the getting-it-all-to-work-together-well-automatically heading, which I said is non-trivial. But Novamente directs a lot of attention to this problems, by among other approaches (a) using long and short term importance metrics to guide computational resource allocation, (b) having a deep memory of which computational patterns have proven appropriate in prior similar circumstances, (c) having a gen/comp hierarchy of such prior computational patterns which allows them to be instantiated in a given case in a context appropriate way, and (d) providing powerful inferencing mechanisms that go way beyond those commonly used in most current AIs. I am totally confident we could get something very useful out of the system even if it was not as well tuned as a human brain. There as all sorts of ways you could dampen the potential not only for combinatorial explosion, but also for instability. We probably would start it out with a lot of such damping, but over time give it more freedom to control its own parameters. RICHARD Are there solutions to the problems of flexible, abstract analogy building? Language learning? ED Not hard! A Novamente class machine would be like Hofstadter’s CopyCat on steroids when it comes to making analogies. The gen/comp hierarchy of patterns would not only apply to all the concepts that fall directly within what we think of as NL, but also to the system’s world-knowledge, itself, of which such NL concepts and their contexts would be a part. This includes knowledge about its own life-history, behavior, and the feedback it has received. Thus, it would be fully capable of representing and matching concepts at the level humans do when understanding and communicating with NL. The deep contextual grounding contained within such world knowledge and the ability to make inferences from it in real time would largely solve the hard disambiguation problems in natural language recognition, and allow language generation to be performed rapidly in a way that is appropriate to all the levels of context that humans use when speaking. RICHARD Pragmatics? ED Not hard! Follows from the above answer. Understanding of pragmatics would result from the ability to dynamically generalize from prior similar statements in prior similar contexts, of what those prior contexts contained. RICHARD Ben Goertzel wrote: Goertzel This practical design is based on a theory that is fairly complete, but not easily verifiable using current technology. The verification, it seems, will come via actually getting the AGI built! ED You and Ben are totally correct. None of this will be proven until it has actually been shown to work. But significant pieces of it have already been shown to work. I think Ben believes it will work, as do I, but we both agree it will not be “verifiable” until it actually does.
Re: [agi] What best evidence for fast AI?
Richard, Thus: if someone wanted volunteers to fly in their brand-new aircraft design, but all they could do to reassure people that it was going to work were the intuitions of suitably trained individuals, then most rational people would refuse to fly - they would want more than intuitions. Yeah, sure. I wouldn't trust the Novamente design's AGI potential, at this stage, nearly enough to allow the life of one of my kids to depend on it. But I trust cars and airplanes in this manner every day. Novamente is a promising-looking RD project, not a proven technology; that's obvious. In this light, my summary would not be a distortion of your position at all, but only a statement about whether an appeal to intuition counts as a good reason to believe. Just to be clear: the whole design doesn't have to be taken in one big gulp of mysterious intuition. There are plenty of well-substantiated aspects, substantiated by math or by prototype experiments or functionalities of various system components. But there are some aspects whose ability to deliver the desired functionality is not yet well substantiated, also. And, of course, there are some suitably trained individuals who do not share your intuitions, even given the limited access they have to your detailed design. So far, no one who has taken the time to carefully study the detailed design has come forward and told me I think that ain't gonna work. Varying levels of confidence have been expressed; and most of all, the opinion has been expressed that the design is complicated and even though the whole thing seems to make a lot of sense, there are a heck of a lot of details to be resolved. I respect your optimism, and applaud your single-minded commitment to the project: if it is going to work, that is the way to get it done. I certainly wish you luck with it. Thanks! Ben - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=64085576-1e462a
Re: [agi] What best evidence for fast AI?
Hi, The following was my brief reply when someone asked me recently why I think AGI is coming: 1. New constructive theories and engineering plans on AGI begin to appear after decades of vacancy on this topic --- AGI won't be possible until someone begin to try 2. All proposed arguments on the impossibility of AGI failed to settle the debate --- if something isn't proven impossible, it remains possible 3. More and more people get disappointed by the mainstream AI research --- if you want AGI, you must directly work on it, not on a piece cut from it arbitrarily 4. The advance of computer techniques, both in hardware and software, make system development much easier --- an individual or a small team can go quite far 5. The Web let the small number of AGI believers speak to and hear from each other, and an AGI community is forming --- not only the widely accepted opinions can be heard 6. Theoretical progress in the related cognitive sciences --- to build AGI, it is needed to know the I in it first As for the rapid progress part of your question, of course it will be considered as rapid, compared to the last two decades, when there wasn't much progress in this direction at all. I don't expect the above answer to convince a wide academic audience --- that requires a much more detailed and technical analysis. To my opinion, even when AGI is finally achieved, it will still take some people some time to acknowledge its intelligence, since it will be very different from their expectation. Pei Wang http://nars.wang.googlepages.com/ On Nov 10, 2007 6:41 AM, Robin Hanson [EMAIL PROTECTED] wrote: I've been invited to write an article for an upcoming special issue of IEEE Spectrum on Singularity, which in this context means rapid and large social change from human-level or higher artificial intelligence. I may be among the most enthusiastic authors in that issue, but even I am somewhat skeptical. Specifically, after ten years as an AI researcher, my inclination has been to see progress as very slow toward an explicitly-coded AI, and so to guess that the whole brain emulation approach would succeed first if, as it seems, that approach becomes feasible within the next century. But I want to try to make sure I've heard the best arguments on the other side, and my impression was that many people here expect more rapid AI progress. So I am here to ask: where are the best analyses arguing the case for rapid (non-emulation) AI progress? I am less interested in the arguments that convince you personally than arguments that can or should convince a wide academic audience. [I also posted this same question to the sl4 list.] Robin Hanson [EMAIL PROTECTED] http://hanson.gmu.edu Research Associate, Future of Humanity Institute at Oxford University Associate Professor of Economics, George Mason University MSN 1D3, Carow Hall, Fairfax VA 22030- 703-993-2326 FAX: 703-993-2323 This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63814140-f1835c
Re: [agi] What best evidence for fast AI?
my inclination has been to see progress as very slow toward an explicitly-coded AI, and so to guess that the whole brain emulation approach would succeed first Why are you not considering a seed/learning AGI? - Original Message - From: Robin Hanson To: agi@v2.listbox.com Sent: Saturday, November 10, 2007 6:41 AM Subject: [agi] What best evidence for fast AI? I've been invited to write an article for an upcoming special issue of IEEE Spectrum on Singularity, which in this context means rapid and large social change from human-level or higher artificial intelligence. I may be among the most enthusiastic authors in that issue, but even I am somewhat skeptical. Specifically, after ten years as an AI researcher, my inclination has been to see progress as very slow toward an explicitly-coded AI, and so to guess that the whole brain emulation approach would succeed first if, as it seems, that approach becomes feasible within the next century. But I want to try to make sure I've heard the best arguments on the other side, and my impression was that many people here expect more rapid AI progress. So I am here to ask: where are the best analyses arguing the case for rapid (non-emulation) AI progress? I am less interested in the arguments that convince you personally than arguments that can or should convince a wide academic audience. [I also posted this same question to the sl4 list.] Robin Hanson [EMAIL PROTECTED] http://hanson.gmu.edu Research Associate, Future of Humanity Institute at Oxford University Associate Professor of Economics, George Mason University MSN 1D3, Carow Hall, Fairfax VA 22030- 703-993-2326 FAX: 703-993-2323 -- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63822575-74b1e4
Re: [agi] What best evidence for fast AI?
At 09:10 AM 11/10/2007, you wrote: my inclination has been to see progress as very slow toward an explicitly-coded AI, and so to guess that the whole brain emulation approach would succeed first Why are you not considering a seed/learning AGI? That would count for non-emulation AI, which is what I intended to ask about. Robin Hanson [EMAIL PROTECTED] http://hanson.gmu.edu Research Associate, Future of Humanity Institute at Oxford University Associate Professor of Economics, George Mason University MSN 1D3, Carow Hall, Fairfax VA 22030- 703-993-2326 FAX: 703-993-2323 This list is sponsored by AGIRI: http://www.agiri.org/emailTo unsubscribe or change your options, please go to:http://v2.listbox.com/member/?member_id=8660244_secret=63826231-f92c46
RE: [agi] What best evidence for fast AI?
Hi Robin. In part it depends on what you mean by fast. 1. Fast - less than 10 years. I do not believe there are any strong arguments for general-purpose AI being developed in this timeframe. The argument here is not that it is likely, but rather that it is *possible*. Some AI researchers, such as Marvin Minsky, believe that we already have the necessary hardware commonly available, if we only knew what software to write for it. If, as seems likely, there is a large economic incentive for the development of this software, it seems reasonable to grant the possibility that it will be developed. Following that line of reasoning, a computation of probability * impact yields a large number for even small probabilities since the impact of a technological singularity could be very large. So planning for the possibility seems prudent. 2. Fast - less than 50 years. For this timeframe, just dust off Moravec's old computer speed chart. On such a chart I think we're supposed to be at something like mouse level right now -- and in fact we have seen supercomputers beginning to take a shot at simulating mouse-brain-like structures. It does not feel so wrong to think that the robot cars succeeding in the DARPA challenges are maybe up to mouse-level capabilities. It is certainly possible that once computers surpass the raw processing power of the human brain by 10, 100, 1000 times, we will just be too stupid to keep up with their capabilities for some reason, but it seems like a more reasonable bet to me that the economic pressures to make somewhat good use of available computing resources will win out. AI is often called a perpetual failure, but from this view that is not true at all; AI has been a spectacular success. It's very impressive that the early researchers were able to get computers with nematode-level nervous systems to show any interesting cognitive behavior at all. At worst, AI is keeping up with the available machine capabilities admirably. Still, putting aside the brain simulation route, we do have to build models of mind that actually work. As Pei Wang just pointed out, we are beginning to see models such as Ben Goertzel's Novamente that at least seem like they might have a shot at sufficiency. That is not proof, but it is an indication that we may not be overmatched by this challenge, once the machinery becomes available. If something like Moore's law continues (I suppose it's a cognitive bias to assume it will continue and a different bias to assume it won't), who wants to bet that computers 10,000, 100,000, or 1,000,000 times as powerful as our brains will go to waste? Add as many zeros as you want... they cost five years each. - Having written that, I confess it is not completely convincing. There are a lot of assumptions involved. I don't think there *is* an objectively convincing argument. That's why I never try to convince anybody... I can play in the intersection between engineering and wishful thinking if I want, simply because it amuses me more than watching football. Hopefully some folks with more earnest beliefs will have better arguments for you. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63831279-12920a
Re: [agi] What best evidence for fast AI?
AGI might turn out to be relatively easy to implement, if right theory comes along, so there's some chance of building AGI in the nearest future, while there's NO chance of implementing brain emulation before all those numerous technical details are tackled, and it can take really long time, which WILL take many lives. -- Vladimir Nesovmailto:[EMAIL PROTECTED] - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63832567-75155c
RE: [agi] What best evidence for fast AI?
At 10:29 AM 11/10/2007, Derek Zahn wrote: 2. Fast - less than 50 years. For this timeframe, just dust off Moravec's old computer speed chart. On such a chart I think we're supposed to be at something like mouse level right now -- and in fact we have seen supercomputers beginning to take a shot at simulating mouse-brain-like structures. It does not feel so wrong to think that the robot cars succeeding in the DARPA challenges are maybe up to mouse-level capabilities. ... AI has been a spectacular success. It's very impressive that the early researchers were able to get computers with nematode-level nervous systems to show any interesting cognitive behavior at all. At worst, AI is keeping up with the available machine capabilities admirably. My impression is that the cognitive performance of mice is vastly superior to that of current robot cars. I don't see how they could be considered even remotely comparable. But perhaps I have misjudged. Has anyone attempted to itemize an inventory of mouse mental abilities, and compared that to current robot abilities? Robin Hanson [EMAIL PROTECTED] http://hanson.gmu.edu Research Associate, Future of Humanity Institute at Oxford University Associate Professor of Economics, George Mason University MSN 1D3, Carow Hall, Fairfax VA 22030- 703-993-2326 FAX: 703-993-2323 This list is sponsored by AGIRI: http://www.agiri.org/emailTo unsubscribe or change your options, please go to:http://v2.listbox.com/member/?member_id=8660244_secret=63834458-b785dd
Re: [agi] What best evidence for fast AI?
On Saturday 10 November 2007 09:29, Derek Zahn wrote: On such a chart I think we're supposed to be at something like mouse level right now -- and in fact we have seen supercomputers beginning to take a shot at simulating mouse-brain-like structures. Ref? - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63834893-c2b731
Re: [agi] What best evidence for fast AI?
On 11/10/07, Bryan Bishop [EMAIL PROTECTED] wrote: On Saturday 10 November 2007 09:29, Derek Zahn wrote: On such a chart I think we're supposed to be at something like mouse level right now -- and in fact we have seen supercomputers beginning to take a shot at simulating mouse-brain-like structures. Ref? http://news.bbc.co.uk/2/hi/technology/6600965.stm Somebody else can probably provide more technical details, as well as information about where this research is now, half a year later. -- http://www.saunalahti.fi/~tspro1/ | http://xuenay.livejournal.com/ Organizations worth your time: http://www.singinst.org/ | http://www.crnano.org/ | http://lifeboat.com/ - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63835208-fffe86
Re: [agi] What best evidence for fast AI?
On Saturday 10 November 2007 10:07, Kaj Sotala wrote: http://news.bbc.co.uk/2/hi/technology/6600965.stm The researchers say that although the simulation shared some similarities with a mouse's mental make-up in terms of nerves and connections it lacked the structures seen in real mice brains. Looks like they were just simulating eight million neurons with up to 6.3k synapses each. How's that necessarily a mouse simulation, anyway? - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63838397-7d08b6
Re: [agi] What best evidence for fast AI?
Looks like they were just simulating eight million neurons with up to 6.3k synapses each. How's that necessarily a mouse simulation, anyway? It really isn't because the individual neuron behavior is so *vastly* simplified. It is, however, a necessary first step and likely to teach us *a lot*. - Original Message - From: Bryan Bishop [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Saturday, November 10, 2007 11:22 AM Subject: Re: [agi] What best evidence for fast AI? On Saturday 10 November 2007 10:07, Kaj Sotala wrote: http://news.bbc.co.uk/2/hi/technology/6600965.stm The researchers say that although the simulation shared some similarities with a mouse's mental make-up in terms of nerves and connections it lacked the structures seen in real mice brains. Looks like they were just simulating eight million neurons with up to 6.3k synapses each. How's that necessarily a mouse simulation, anyway? - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63842006-af107f
RE: [agi] What best evidence for fast AI?
Bryan Bishop: Looks like they were just simulating eight million neurons with up to 6.3k synapses each. How's that necessarily a mouse simulation, anyway? It isn't. Nobody said it was necessarily a mouse simulation. I said it was a simulation of a mouse-brain-like structure. Unfortunately, not enough is yet known about specific connectivity so the best that can be done is play with structures of similar scale in anticipation of further advances. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63847412-4e7cf3
Re: [agi] What best evidence for fast AI?
On Saturday 10 November 2007 11:31, Derek Zahn wrote: Unfortunately, not enough is yet known about specific connectivity so the best that can be done is play with structures of similar scale in anticipation of further advances. What signs will tell us that we do know enough about the architecture of the mouse brain to simulate it to some degree of usefulness? - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63855916-7f88e6
Re: [agi] What best evidence for fast AI?
On 11/10/07, Robin Hanson [EMAIL PROTECTED] wrote: My impression is that the cognitive performance of mice is vastly superior to that of current robot cars. I don't see how they could be considered even remotely comparable. But perhaps I have misjudged. Has anyone attempted to itemize an inventory of mouse mental abilities, and compared that to current robot abilities? It might be worthwhile to point out robotic technology is currently on a rapidly advancing segment of the curve, exploiting low-hanging fruit recently reachable by a convergence of capabilities becoming affordable including significant processing power, memory, batteries, wireless comm, motors and actuators, etc. In my opinion, the availability of the hardware is defining the near-term potential, with competition accelerating the rush to fill that void. Development beyond that level, however, proceeds at a much slower evolutionary rate. Much as natural language processing made substantial gains and then leveled off distinctly below the level of human understanding, robotics development is accelerating toward that level at which the rate of progress will sharply plateau. At the DARPA Urban Challenge last weekend, the optimism and flush of rapid growth was palpable, but as I was driving home I approached a truck off the side of the road, its driver pulling hard on a bar, tightening the straps securing the load. Without conscious thought I moved over in my lane to allow for the possibility that he might slip. That chain of inference, and its requisite knowledge base, leading to a simple human behavior, are not even on the radar horizon of current AI technology. - Jef - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63856239-c9c2f5
Re: [agi] What best evidence for fast AI?
I think the media coverage of mouse brain simulation was a little misleading. What I think they actually achieved was to simulate many neurons based upon the Izhikevich model on a large computer at a rate significantly slower than real time. As far as I know there was no attempt to actually simulate the brain structures of a mammal. On 10/11/2007, Kaj Sotala [EMAIL PROTECTED] wrote: http://news.bbc.co.uk/2/hi/technology/6600965.stm Somebody else can probably provide more technical details, as well as information about where this research is now, half a year later. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63857903-1e305f
RE: [agi] What best evidence for fast AI?
Robin, I am an evangelist for the fact that the time for powerful AI could be here very rapidly if there were reasonable funding for the right people. There is a small, but increasing number of people who pretty much understand how to build artificial brains as powerful as that of humans, not 100% but probably at least 90% at an architectual level. What is needed is funding. It will come, but exactly how fast, and to which people, is the big question. The below paper is written with the assumption that someone -- some VC's, Governments, Google, Microsoft, Intel, some Chinese multi-billionaire -- makes a significant investment in the right people. I have cobbled this together rapidly from some similar prior writings, so please forgive the typos. I assume you will only pick through it for ideas, so exact language is not important. I you have any questions, please call or email me. Ed Porter == The Time for Powerful General AI is Rapidly Approaching by Edward Porter The time for powerful general AI is rapidly approaching. Its beginnings could be here in two to ten years if the right people got the right funding. Starting in two years it could start providing the first in a series of ever-more-powerful, ever-more-valuable, market-dominating products. In five to ten years it could be delivering true superhuman intelligence. In that time frame, for example, this would enable software running on less than $3 million dollar hardware to write reliable code faster than a thousand human programmers - or, with a memory swap, to remember every word, every concept, every stated rational in a world-class law library and to reason from that knowledge hundreds to millions of times faster than a human lawyer, depending on the exact nature of the reasoning task. You should be skeptical. The AI field has been littered with false claims before. But for each of history's long-sought, but long-delayed, technical breakthroughs, there has always come a time when it finally happened. There is strong reason to believe that for powerful machine intelligence that time is now. What is the evidence? It has two major threads. The first is that for the first time in history we have hardware with the computational power to support near-human intelligence, and in five to seven years the cost of hardware powerful enough to support superhuman intelligence could be as low as $200,000 to $3,000,000, meaning that virtually every medium to mid-size organization will want many of them. The second is that, due to advances in brain science and in AI, itself, there are starting to be people, like those at Novamente LLC, who have developed reasonable and detailed architectures for how to use such powerful hardware efficiently to create near- or super-human intelligence. THE HARDWARE To do computation of the general type and sophistication of the human brain, you need something within at least several orders of magnitude of the capacity of the human brain, itself, in each of three dimensions: representational, computational, and intercommunication capacity. You can't have the common sense, intuition, and context appropriateness of a human mind unless you can represent and rapidly make generalizations from and inference between substantially all parts of world knowledge - where world knowledge is the name given to the extremely large body of experientially derived knowledge most humans have. Most past AI work has been done on machines that have less than one one millionth the capacity in one or more of these three dimensions. This is like trying to do what the human brain does with a brain roughly 2000 times smaller than that of a rat. No wonder most prior attempts at human-level AI have had so many false promises and failures. No wonder the correct, large-hardware approaches have been up until very recently impossible to properly demonstrate and, thus, get funding for. And, thus, no wonder, the AI establishment does not understand such correct approaches. But human-level hardware is coming soon. Systems are already available for under ten million dollars (with roughly 4.5K 2Ghz 4 core processors, 168 TeraFlops/sec, a nominal bandwidth of 4TBytes/sec, and massive hard disk storage) that are very roughly human level in two out of the above three dimensions. These machines are very roughly 1000 times slower than humans with regard to messaging interconnect, but they are also hundreds of millions of times faster than humans for many of the tasks at which machines already out perform us. Even machines with much less hardware could provide marketable powerful intelligences. AIs that were substantially sub-human at some tasks could combine that sub-human intelligence with the skill at which computers greatly out perform us to produce combined intelligences that could be extremely valuable for many tasks. Furthermore,
Re: [agi] What best evidence for fast AI?
On 10/11/2007, Jef Allbright [EMAIL PROTECTED] wrote: At the DARPA Urban Challenge last weekend, the optimism and flush of rapid growth was palpable, but as I was driving home I approached a truck off the side of the road, its driver pulling hard on a bar, tightening the straps securing the load. Without conscious thought I moved over in my lane to allow for the possibility that he might slip. That chain of inference, and its requisite knowledge base, leading to a simple human behavior, are not even on the radar horizon of current AI technology. I was saying to someone recently that it's hard to watch something like the recent Urban Challenge and argue convincingly that AI is not making progress or that it's been a failure. Admittedly the intelligence here is not smart enough to carry out the sort of reasoning you describe, such as I see a large object and predict that it may be about to fall so I better move out of the way. However, the path to this sort of ability just involves more accurate 3D modelling of the environment together with intelligent segmentation and some naive physics applied. It's the perception accuracy/modeling which is key to being able to implement these skills, which a mouse may or may not be capable of (I don't know enough about the cognitive skills of mice to be able to say). - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63859193-c54189
Re: [agi] What best evidence for fast AI?
On Saturday 10 November 2007 12:52, Edward W. Porter wrote: In fact, if the ITRS roadmap projections continue to be met through What is the ITRS roadmap? Do you have a link? - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63859781-dcb1eb
Re: [agi] What best evidence for fast AI?
Robin Hanson wrote: I've been invited to write an article for an upcoming special issue of IEEE Spectrum on Singularity, which in this context means rapid and large social change from human-level or higher artificial intelligence. I may be among the most enthusiastic authors in that issue, but even I am somewhat skeptical. Specifically, after ten years as an AI researcher, my inclination has been to see progress as very slow toward an explicitly-coded AI, and so to guess that the whole brain emulation approach would succeed first if, as it seems, that approach becomes feasible within the next century. But I want to try to make sure I've heard the best arguments on the other side, and my impression was that many people here expect more rapid AI progress. So I am here to ask: where are the best analyses arguing the case for rapid (non-emulation) AI progress? I am less interested in the arguments that convince you personally than arguments that can or should convince a wide academic audience. I gave my answer to this question in a paper I presented at the 2006 AGIRI workshop on Artificial General Intelligence [1]. Stripped to its core, the argument is that AI progress has been slow for a specific reason, not because the problem is intrinsically hard. The reason for the slow progress is a fundamental misperception of the nature of the AI problem: intelligent systems (by which I mean completely general intelligent systems that are capable of acquiring knowledge on their own initiative) *probably* contain an irreducible element of complexity, in the 'complex systems' sense of 'complexity'. The two main consequence of this complexity are that (1) we would expect some of an AI's low-level mechanisms to have an opaque relationship to the AI's overall behavior (i.e. there are mechanisms down there that do not look like they have any bearing whatsoever on the intelligence of the overall system, and yet they play an indispensible role in the system's intelligent performance), and (2) the only way to get around the problems caused by (1) would be to make a systematic effort to emulate the human cognitive system -- not at the neural level, mark you, but at the cognitive level. The final conclusion of the argument I give in the paper is an interesting sociology-of-science observation that bears directly on your question of how rapidly we could get to full AGI: unfortunately, the AI community is populated with people who have an extremely strong bias against accepting these arguments, and this strong bias is what is holding back progress. Basically, 'traditional' AI people have an almost theological aversion to the idea that the task of building an AI might involve having to learn (and deconstruct!) a vast amount of cognitive science, and then use an experimental-science methodology to find the mechanisms that really give rise to AI. AI people are, at heart, mathematicians, and this is serious problem if the only way to succeed has little to do with mathematics. Looked at in this way, the answer to your question is that if a new type of AI comes along (what I have dubbed 'theoretical psychology' because of its unique relationship to AI and psychology) and if it gathers enough support, we could find that the progress rate of this new approach bears no relationship to the progress rate of AI over the last fifty years. I have started the process of building the infrastructure needed to do this kind of work. So far this is working well: among other things, a colleague of mine (Trevor Harley) and I have started re-analyzing the literature of cognitive science to bring it into line with the new approach, and our efforts have met with some surprising early successes (the first fruits of this effort being a cognitive neuroscience paper that is currently in press [2]). From my point of view, old-style cognitive science and old-style AI are both falling neatly and elegantly into this new framework, so my personal feeling is that a new period of rapid progress is just over the horizon, and that human-level AGI might happen in the coming decade. If it were not for this particular way of seeing the problems of AI, I would be with the skeptics: I think that conventional AI will not yield a singularity-class AGI for a long time (if ever), and I believe that the brain-emulation folks are being wildly optimistic about what they can achieve, because they are blind to functional-level issues, and do not have the resolution or in-vivo tools needed to reach their goals. Regards Richard Loosemore References. [1] Loosemore, R.P.W. (2007). Complex Systems, Artificial Intelligence and Theoretical Psychology. In B. Goertzel P. Wang, Proceedings of the 2006 AGI Workshop. Amsterdam: IOS Press. This can be found online at http://www.agiri.org/wiki/Workshop_Proceedings (chapter 11). [2] Loosemore, R.P.W. Harley, T.A. Brains and Minds: On
Re: [agi] What best evidence for fast AI?
On 11/10/07, Edward W. Porter [EMAIL PROTECTED] wrote: There is a small, but increasing number of people who pretty much understand how to build artificial brains as powerful as that of humans, not 100% but probably at least 90% at an architectual level. Being 90% certain of where to get on the path is quite different from being 90% certain of the path. - Jef - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63864457-e6c274
RE: [agi] What best evidence for fast AI?
http://www.itrs.net/reports.html Edward W. Porter Porter Associates 24 String Bridge S12 Exeter, NH 03833 (617) 494-1722 Fax (617) 494-1822 [EMAIL PROTECTED] -Original Message- From: Bryan Bishop [mailto:[EMAIL PROTECTED] Sent: Saturday, November 10, 2007 2:03 PM To: agi@v2.listbox.com Subject: Re: [agi] What best evidence for fast AI? On Saturday 10 November 2007 12:52, Edward W. Porter wrote: In fact, if the ITRS roadmap projections continue to be met through What is the ITRS roadmap? Do you have a link? - Bryan - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244id_secret=63868500-801939