Re: [agi] The Effect of Application of an Idea
On Tue, Mar 25, 2008 at 7:19 PM, Vladimir Nesov [EMAIL PROTECTED] wrote: Certainly ambiguity (=applicability to multiple contexts in different ways) and presence of rich structure in presumably simple 'ideas', as you call it, is a known issue. Even interaction between concept clouds evoked by a pair of words is a nontrivial process (triangular lightbulb). In a way, whole operation can be modeled by such interactions, where sensory input/recall is taken to present a stream of triggers that evoke concept cloud after cloud, with associations and compound concepts forming at the overlaps. But of course it's too hand-wavy without a more restricted model of what's going on. Communicating something that exists solely on high level is very inefficient, plus most of such content can turn out to be wrong. Back to prototyping... -- Vladimir Nesov [EMAIL PROTECTED] I agreed with you up until your conclusion. While the problems that I talked about may be known issues, they are discussed almost exclusively using intuitive models, like we used, or by referring to ineffective models, like network theories that do not explicitly show how its associative interrelations would effectively deal with the intricate conceptual details that would be required to address these issues and would be produced by an effective solution. I have never seen any theory that was designed to specifically address the range of situations that I am thinking of although most earlier AI models were intended to deal similar issues and I have seen some exemplary models that did use controlled models which showed how some of these interrelations might be modeled. These intuitive discussions and the exaggerated effectiveness of inadequate programs creates a concept cloud itself, and the problem is that the knowledgeable listener has a feeling that he understands the problem even without having made any kind of commitment to the exploration of an effective solution. Although I have not detailed how the effects of the application of ideas might be modeled in an actual AI program (or in an extremely simple model that I would use to start studying the modeling) my whole point is that if you are interested in advancing AI programming, then the issue that my theory addresses is a problem that can not be dismissed with a wave of the hand. The next step for me is to find a model that would be strong enough to hold up to genuine extensible learning. If you are making a decision on how much time you should spend thinking about this based only on whether or not you have thought about similar problems I believe that you have already considered some sampling of the kind of problems that my theory is meant to address. Jim Bromer --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
On Wed, Mar 26, 2008 at 4:27 PM, Jim Bromer [EMAIL PROTECTED] wrote: I agreed with you up until your conclusion. While the problems that I talked about may be known issues, they are discussed almost exclusively using intuitive models, like we used, or by referring to ineffective models, like network theories that do not explicitly show how its associative interrelations would effectively deal with the intricate conceptual details that would be required to address these issues and would be produced by an effective solution. I have never seen any theory that was designed to specifically address the range of situations that I am thinking of although most earlier AI models were intended to deal similar issues and I have seen some exemplary models that did use controlled models which showed how some of these interrelations might be modeled. These intuitive discussions and the exaggerated effectiveness of inadequate programs creates a concept cloud itself, and the problem is that the knowledgeable listener has a feeling that he understands the problem even without having made any kind of commitment to the exploration of an effective solution. Although I have not detailed how the effects of the application of ideas might be modeled in an actual AI program (or in an extremely simple model that I would use to start studying the modeling) my whole point is that if you are interested in advancing AI programming, then the issue that my theory addresses is a problem that can not be dismissed with a wave of the hand. The next step for me is to find a model that would be strong enough to hold up to genuine extensible learning. If you are making a decision on how much time you should spend thinking about this based only on whether or not you have thought about similar problems I believe that you have already considered some sampling of the kind of problems that my theory is meant to address. What you describe is essentially my own path up to this point: I started with considering high-level capabilities and gradually worked towards an implementation that seems to be able to exhibit these high-level capabilities. At the end of my last message I referred to a pragmatic problem. Substrate with which I now experiment is essentially a very simple recurrent network with seemingly insignificant tweaks. Without high-level view of how to make it exhibit high-level capabilities I'd never look at it twice. Convincing someone else that it is that capable will take a rather long description, and I can well turn out to be wrong (so people have a perfectly good reason not to listen). It seems more sensible to stick to prototyping and wait for more solid results, either changing the theory, or demonstrating its potential. -- Vladimir Nesov [EMAIL PROTECTED] --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
On 25/03/2008, Vladimir Nesov [EMAIL PROTECTED] wrote Simple systems can be computationally universal, so it's not an issue in itself. On the other hand, no learning algorithm is universal, there are always distributions that given algorithms will learn miserably. The problem is to find a learning algorithm/representation that has the right kind of bias to implement human-like performance. First a riddle: What can be all learning algorithms, but is none? I'd disagree. Okay simple systems can be computationally universal, but what does that really mean. Computational universality means to be able to represent any computable function, the range and domain of this function are assumed to be from the natural numbers to itself. Most AI formulations when they say that are computationally universal are only talking about function of F: I → O where I is the input and O is the output. These include the formulations of neural networks/GA etc that I have seen. However there are lots of interesting programs in computers that do not map the input to the output. Humans also do not just map the input to the output, we also think, ruminate, model and remember. This does not affect the range of functions from the input to the output, but it does change how quickly they can be moved between. What I am interested in is in systems where the ranges and domains of the functions are entities inside the system. That is the F: I → S, F: S → O, and F: S→ S are important and should be potentially computationally universal. Where S is the internal memory of the system. This allows the system to be all possible learning algorithms (although only one at any time), but also it is no algorithm (else F: I x S → S, would be fixed). General purpose desktop computers are these kinds of systems. If they weren't how else could we implement any type of learning system on them? Thus the answer to my riddle. The question I have been trying to answer precisely is how to govern these sorts of systems so they roughly do what you want, without you having to give precise instructions. Will Pearson --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
On Wed, Mar 26, 2008 at 10:17 AM, Vladimir Nesov [EMAIL PROTECTED] wrote: What you describe is essentially my own path up to this point: I started with considering high-level capabilities and gradually worked towards an implementation that seems to be able to exhibit these high-level capabilities. At the end of my last message I referred to a pragmatic problem. Substrate with which I now experiment is essentially a very simple recurrent network with seemingly insignificant tweaks. Without high-level view of how to make it exhibit high-level capabilities I'd never look at it twice. Convincing someone else that it is that capable will take a rather long description, and I can well turn out to be wrong (so people have a perfectly good reason not to listen). It seems more sensible to stick to prototyping and wait for more solid results, either changing the theory, or demonstrating its potential. -- Vladimir Nesov I do not know much about neural networks, but from what I read, I always felt that a recurrent network would be the only way you could feasibly get an ANN to represent (excuse my french) distinct items without absurdly huge and noisy expansions. So I am curious about what you are talking about. When you mention prototyping, you are talking about prototyping the neural network with high level concepts for easier demonstrations or something like that. I think there was some discussion about using 'labels' in neural networks on one of those links to an online video that were recently posted. Is this similar to what you mean by prototyping? Jim Bromer --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
2008/3/26 William Pearson [EMAIL PROTECTED]: On 25/03/2008, Vladimir Nesov [EMAIL PROTECTED] wrote Simple systems can be computationally universal, so it's not an issue in itself. On the other hand, no learning algorithm is universal, there are always distributions that given algorithms will learn miserably. The problem is to find a learning algorithm/representation that has the right kind of bias to implement human-like performance. First a riddle: What can be all learning algorithms, but is none? Excellent philosophical point!! Okay simple systems can be computationally universal, but what does that really mean. Computational universality means to be able to represent any computable function, the range and domain of this function are assumed to be from the natural numbers to itself. ---I think Godel would disagree. Most AI formulations when they say that are computationally universal are only talking about function of F: I → O where I is the input and O is the output. These include the formulations of neural networks/GA etc that I have seen. However there are lots of interesting programs in computers that do not map the input to the output. Humans also do not just map the input to the output, we also think, ruminate, model and remember. This does not affect the range of functions from the input to the output, but it does change how quickly they can be moved between. What I am interested in is in systems where the ranges and domains of the functions are entities inside the system. That is the F: I → S, F: S → O, and F: S→ S are important and should be potentially computationally universal. Where S is the internal memory of the system. This allows the system to be all possible learning algorithms (although only one at any time), but also it is no algorithm (else F: I x S → S, would be fixed). General purpose desktop computers are these kinds of systems. If they weren't how else could we implement any type of learning system on them? Thus the answer to my riddle. The question I have been trying to answer precisely is how to govern these sorts of systems so they roughly do what you want, without you having to give precise instructions. Will Pearson -I am going to read this more carefully later. However, the first part of the answer to your last question is that the governance of these kinds of systems will be based on general rules (or methods of generality) so you do not need to define all the precise instructions that would be needed. But, there is not one level of universality, there are potentially infinite levels of generalization, and they do not all mesh together perfectly. Although this kind of talk may not solve the problem, I believe that this is where we are going to end up working if we continue to work on the problem Jim Bromer --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
On Wed, Mar 26, 2008 at 8:47 PM, Jim Bromer [EMAIL PROTECTED] wrote: I do not know much about neural networks, but from what I read, I always felt that a recurrent network would be the only way you could feasibly get an ANN to represent (excuse my french) distinct items without absurdly huge and noisy expansions. So I am curious about what you are talking about. When you mention prototyping, you are talking about prototyping the neural network with high level concepts for easier demonstrations or something like that. I think there was some discussion about using 'labels' in neural networks on one of those links to an online video that were recently posted. Is this similar to what you mean by prototyping? For now objective is to try to achieve basic high-level dynamics that this architecture was designed to implement, and thus to partially establish consistence of many-faceted high-level design with simple network implementation. If this stage succeeds, after a bit of scalability implementation it should be possible to start teaching it more impressive high-level feats. -- Vladimir Nesov [EMAIL PROTECTED] --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
I know, I KNOW :-) WAS Re: [agi] The Effect of Application of an Idea
First a riddle: What can be all learning algorithms, but is none? A human being! --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: I know, I KNOW :-) WAS Re: [agi] The Effect of Application of an Idea
On 26/03/2008, Mark Waser [EMAIL PROTECTED] wrote: First a riddle: What can be all learning algorithms, but is none? A human being! Well my answer was a common PC, which I hope is more illuminating because we know it well. But human being works, as does any future AI design, as far as I am concerned. Will Pearson --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
On 24/03/2008, Jim Bromer [EMAIL PROTECTED] wrote: To try to understand what I am talking about, start by imagining a simulation of some physical operation, like a part of a complex factory in a Sim City kind of game. In this kind of high-level model no one would ever imagine all of the objects should interact in one stereotypical way, different objects would interact with other objects in different kinds of ways. And no one would imagine that the machines that operated on other objects in the simulation were not also objects in their own right. For instance the machines used in production might require the use of other machines to fix or enhance them. And the machines might produce or operate on objects that were themselves machines. When you think about a simulation of some complicated physical systems it becomes very obvious that different kinds of objects can have different effects on other objects. And yet, when it comes to AI, people go on an on about systems that totally disregard this seemingly obvious divergence of effect that is so typical of nature. Instead most theories see insight as if it could be funneled through some narrow rational system or other less rational field operations where the objects of the operations are only seen as the ineffective object of the pre-defined operations of the program. How would this differ from the sorts of computational systems I have been muttering about? Where you have an architecture where an active bit of code or program is equivalent to an object in the above paragraph. Also have a look at Eurisko by Doug Lenat. Will Pearson --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
On Tue, Mar 25, 2008 at 11:23 AM, William Pearson [EMAIL PROTECTED] wrote: On 24/03/2008, Jim Bromer [EMAIL PROTECTED] wrote: To try to understand what I am talking about, start by imagining a simulation of some physical operation, like a part of a complex factory in a Sim City kind of game. In this kind of high-level model no one would ever imagine all of the objects should interact in one stereotypical way, different objects would interact with other objects in different kinds of ways. And no one would imagine that the machines that operated on other objects in the simulation were not also objects in their own right. For instance the machines used in production might require the use of other machines to fix or enhance them. And the machines might produce or operate on objects that were themselves machines. When you think about a simulation of some complicated physical systems it becomes very obvious that different kinds of objects can have different effects on other objects. And yet, when it comes to AI, people go on an on about systems that totally disregard this seemingly obvious divergence of effect that is so typical of nature. Instead most theories see insight as if it could be funneled through some narrow rational system or other less rational field operations where the objects of the operations are only seen as the ineffective object of the pre-defined operations of the program. How would this differ from the sorts of computational systems I have been muttering about? Where you have an architecture where an active bit of code or program is equivalent to an object in the above paragraph. Also have a look at Eurisko by Doug Lenat. Will Pearson There is no reason to believe that anything I might imagine would be the same as something that was created 35 years ago! I have a lot of trouble explaining myself on some days. The idea of the effect of the application of ideas is that most people do not consciously think about the subject, and so, just by becoming aware of it one can change how his program works regardless of how automated the program is. It can work with strictly defined logical systems or with inductive systems that can be extended creatively or with systems that are capable of learning. However, it is not a complete solution to AI, it is more like something that you will need to think about if you plan to write some seriously innovative AI application in the near future. So, I haven't written such a program, but I do have something to say. A system that has heuristics that can modify the heuristics of the system is important, and such a system does implement what I am talking about. However, the point is, that Lenat never seemed to completely accept the range that such a thing would have to have to generate true intelligence. The reason is that it would become so complicated that it would make any feasible AI program impossible. And the reason that a truely intelligent AI program is still not feasible is just because it would be complicated. I am saying that the method of recognizing and defining the effect of ideas on other ideas would not, by itself, make it all work, but rather it would help us to better understand how to better automate the kind of extensive complications of effect that would be necessary. I am thinking of a writing about a simple imaginary model that could be incrementally extended. This model would not be useful, because it would be too simple. But I should be able to give you some idea about what I am thinking about. As any program becomes more and more complicated, the programmer has to think more and more about how various combinations of data and processes will interact. Why would anyone think that an advanced AI program would be any simpler? Ideas affect other ideas. Heuristics that can act on other heuristics is a basis of this kind of thing, but it has to be much more complicated than that. So while I don't have the answers, I can begin to think of hand crafting a model where such a thing could be examined, by recognizing that the application of ideas to other ideas will have complicated effects that need to be defined. The more automated AI program would have to use some systems to shape these complicated interactions, but the effect of those heuristics would be modifiable by other learning (to some extent.) Jim Bromer --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
On Tue, Mar 25, 2008 at 2:17 AM, Jim Bromer [EMAIL PROTECTED] wrote: A usage evaluation could be taken as an example of an effect of application, because the idea of usage and of statistical evaluation can be combined with the object of consideration along with other theories that detail how such combinations could be usefully applied to some problem. But it is obviously not the only effective process that would be necessary to understand complicated systems. No one would only use statistical models to discuss the management and operations of a real factory for example. It is rather obvious that such limited methods would be grossly inadequate. Why would anyone imagine that a narrow operational system would be adequate for an AI program? The theory of the effect of application of an idea tries to address this inadequacy by challenging the programmer to begin to think about and program applications that can detail how simple interactive effects can be combined with novel insights in a feasible extensible object. So while I don't have the solution, I believe I can see a path. Simple systems can be computationally universal, so it's not an issue in itself. On the other hand, no learning algorithm is universal, there are always distributions that given algorithms will learn miserably. The problem is to find a learning algorithm/representation that has the right kind of bias to implement human-like performance. It's more or less clear that such representation needs to have higher-level concepts that refine interactions between lower-level concepts and are learned incrementally, built on existing concepts. Association-like processes can port existing high-level circuits to novel tasks for which they were not originally learned, which allows some measure of general knowledge. As I see it, the issue you are trying to solve is the porting of structured high-level competencies. Which looks equivalent to the general problem of association-building between structured representations. Is it roughly a correct characterization of what you are talking about? -- Vladimir Nesov [EMAIL PROTECTED] --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
On Tue, Mar 25, 2008 at 11:30 PM, Jim Bromer [EMAIL PROTECTED] wrote: I am saying that the method of recognizing and defining the effect of ideas on other ideas would not, by itself, make it all work, but rather it would help us to better understand how to better automate the kind of extensive complications of effect that would be necessary. It's interesting, but first the structure of 'ideas' needs to be described, otherwise it doesn't help. As any program becomes more and more complicated, the programmer has to think more and more about how various combinations of data and processes will interact. Why would anyone think that an advanced AI program would be any simpler? Ideas affect other ideas. Heuristics that can act on other heuristics is a basis of this kind of thing, but it has to be much more complicated than that. So while I don't have the answers, I can begin to think of hand crafting a model where such a thing could be examined, by recognizing that the application of ideas to other ideas will have complicated effects that need to be defined. The more automated AI program would have to use some systems to shape these complicated interactions, but the effect of those heuristics would be modifiable by other learning (to some extent.) Modularity fights this problem in programming, helping to keep track of *code*. But this code is built on top of existing models of program's behavior existing in programmers' minds. Programmers manually determine applicability of code. It's often possible to solve a wide variety of problems with existing codebase, but programmer is needed to contextually match and assemble pathways that solve any given problem. We don't currently have practically applicable methods to extend the context in which code can be applied, and to build on these extended contexts. I think that one of the most important features of AGI system must be automated extensibility. It should be possible to teach it new things without breaking it. It should be able to correct its performance to preserve previously learned skills, so that teaching needs only to focus on few high-level performance properties, regardless on how much is already learned. -- Vladimir Nesov [EMAIL PROTECTED] --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
On Tue, Mar 25, 2008 at 4:42 PM, Vladimir Nesov [EMAIL PROTECTED] wrote: Simple systems can be computationally universal, so it's not an issue in itself. On the other hand, no learning algorithm is universal, there are always distributions that given algorithms will learn miserably. The problem is to find a learning algorithm/representation that has the right kind of bias to implement human-like performance. It's more or less clear that such representation needs to have higher-level concepts that refine interactions between lower-level concepts and are learned incrementally, built on existing concepts. Association-like processes can port existing high-level circuits to novel tasks for which they were not originally learned, which allows some measure of general knowledge. As I see it, the issue you are trying to solve is the porting of structured high-level competencies. Which looks equivalent to the general problem of association-building between structured representations. Is it roughly a correct characterization of what you are talking about? Vladimir Nesov [EMAIL PROTECTED] Can you give some more indication about what you mean by porting of structured high-level competencies and the problem of association-building between structured representations? I do not know where you got the phrase porting from since I have only seen it in reference to porting code from one machine to another. I assume that you are using it as a kind of metaphor, or the application of an idea very similar to 'porting' to AGI. Let's suppose that I claim that Ed bumped into me. Right away we can see that the word-concept bumped has some effect on any ideas you might have about Ed, me and Ed and me. My claim here is that the effect of the interaction of ideas goes beyond semantics into the realm of ideas proper. If it turned out that I got into Ed's way (perhaps intentionally) then one might wonder if the claim that Ed bumped into me was a correct or adequate description of what happened. On the other hand, such detail might not be interesting or necessary in some other conversation, so the effect of the idea of 'bumping' and the idea of 'getting in the way of' may or may not be of interest in all conversatations about the event. Furthermore, the idea of 'getting in the way of' may not be relevant to some examinations of what happened, as in the case where a judge might want to focus on whether or not the bumping actually took place. From this kind of focus, the question of whether or not I got in Ed's way might then become evidence of whether or not the bump actually took place, but it would not otherwise be relevant to the judge's examination of the incident. Presentations like the one that I just made have been made often before. What I am saying is that the effect of the application of different ideas may be more clearly deliniated in stories like this, and that process can be seen as a generalization of form that may be used with representations to help show what kind of structure would be needed to create and maintain such complexes of potential relations between ideas. While I do not know the details of how I might go about to create a program to build structure like that, the view that it is only a 'porting of structure' implies that the method might be applied in some simple manner. While it can be applied in a simple manner to a simple model, my interest in the idea is that I could also take the idea further in more complicated models. The point that the method can be used in a simplistic, constrained model is significant because the potential problem is so complex that constrained models may be used to study details that would be impossible in more dynamic learning models. Jim Bromer --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
On Wed, Mar 26, 2008 at 1:27 AM, Jim Bromer [EMAIL PROTECTED] wrote: Let's suppose that I claim that Ed bumped into me. Right away we can see that the word-concept bumped has some effect on any ideas you might have about Ed, me and Ed and me. My claim here is that the effect of the interaction of ideas goes beyond semantics into the realm of ideas proper. If it turned out that I got into Ed's way (perhaps intentionally) then one might wonder if the claim that Ed bumped into me was a correct or adequate description of what happened. On the other hand, such detail might not be interesting or necessary in some other conversation, so the effect of the idea of 'bumping' and the idea of 'getting in the way of' may or may not be of interest in all conversatations about the event. Furthermore, the idea of 'getting in the way of' may not be relevant to some examinations of what happened, as in the case where a judge might want to focus on whether or not the bumping actually took place. From this kind of focus, the question of whether or not I got in Ed's way might then become evidence of whether or not the bump actually took place, but it would not otherwise be relevant to the judge's examination of the incident. Presentations like the one that I just made have been made often before. What I am saying is that the effect of the application of different ideas may be more clearly deliniated in stories like this, and that process can be seen as a generalization of form that may be used with representations to help show what kind of structure would be needed to create and maintain such complexes of potential relations between ideas. While I do not know the details of how I might go about to create a program to build structure like that, the view that it is only a 'porting of structure' implies that the method might be applied in some simple manner. While it can be applied in a simple manner to a simple model, my interest in the idea is that I could also take the idea further in more complicated models. The point that the method can be used in a simplistic, constrained model is significant because the potential problem is so complex that constrained models may be used to study details that would be impossible in more dynamic learning models. Certainly ambiguity (=applicability to multiple contexts in different ways) and presence of rich structure in presumably simple 'ideas', as you call it, is a known issue. Even interaction between concept clouds evoked by a pair of words is a nontrivial process (triangular lightbulb). In a way, whole operation can be modeled by such interactions, where sensory input/recall is taken to present a stream of triggers that evoke concept cloud after cloud, with associations and compound concepts forming at the overlaps. But of course it's too hand-wavy without a more restricted model of what's going on. Communicating something that exists solely on high level is very inefficient, plus most of such content can turn out to be wrong. Back to prototyping... -- Vladimir Nesov [EMAIL PROTECTED] --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
Thanks for asking. I will try to come up with a simple model during the next week. I can create an example because the principle can be used in well-defined constrained models or in more extensible models. The theory does not answer all questions about AGI. I would think that should be taken as a reasonable assumption about any single theory; however, I believe that it can help in the discovery of more dynamic and flexible principles that may be of some use in AI. The reason I think this is because the theory explicitly deals with an issue that has not been abstracted and highlighted in any discussions that I can recall. So while the idea of an effect of application of an idea may have been implicitly invoked in any number of discussions, I don't really think that it has ever really emerged as a fundamental subject matter in its own right. Concept grounding could be taken as an example of the effect of application. The association of a concept with some data that exemplifies it within the greater data environment would naturally produce some kinds of knowledge that could affect other kinds of knowledge. To try to understand what I am talking about, start by imagining a simulation of some physical operation, like a part of a complex factory in a Sim City kind of game. In this kind of high-level model no one would ever imagine all of the objects should interact in one stereotypical way, different objects would interact with other objects in different kinds of ways. And no one would imagine that the machines that operated on other objects in the simulation were not also objects in their own right. For instance the machines used in production might require the use of other machines to fix or enhance them. And the machines might produce or operate on objects that were themselves machines. When you think about a simulation of some complicated physical systems it becomes very obvious that different kinds of objects can have different effects on other objects. And yet, when it comes to AI, people go on an on about systems that totally disregard this seemingly obvious divergence of effect that is so typical of nature. Instead most theories see insight as if it could be funneled through some narrow rational system or other less rational field operations where the objects of the operations are only seen as the ineffective object of the pre-defined operations of the program. A usage evaluation could be taken as an example of an effect of application, because the idea of usage and of statistical evaluation can be combined with the object of consideration along with other theories that detail how such combinations could be usefully applied to some problem. But it is obviously not the only effective process that would be necessary to understand complicated systems. No one would only use statistical models to discuss the management and operations of a real factory for example. It is rather obvious that such limited methods would be grossly inadequate. Why would anyone imagine that a narrow operational system would be adequate for an AI program? The theory of the effect of application of an idea tries to address this inadequacy by challenging the programmer to begin to think about and program applications that can detail how simple interactive effects can be combined with novel insights in a feasible extensible object. So while I don't have the solution, I believe I can see a path. I feel that by using the principle of the effect of the application of ideas, one could build simple extensible models. The models would start out as being simplistic. But by carefully studying how complicated interactions interfere or cohere I believe that some new AI principles may be found. I will try to come up with a simple model during the next week. Jim Bromer On Sun, Mar 23, 2008 at 4:53 AM, Vladimir Nesov [EMAIL PROTECTED] wrote: Jim, It sounds like something about concept grounding, but that's all I got. Can you give an example that demonstrates the structure of what you are talking about? -- Vladimir Nesov [EMAIL PROTECTED] --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
Re: [agi] The Effect of Application of an Idea
Jim, It sounds like something about concept grounding, but that's all I got. Can you give an example that demonstrates the structure of what you are talking about? -- Vladimir Nesov [EMAIL PROTECTED] --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244id_secret=98558129-0bdb63 Powered by Listbox: http://www.listbox.com
[agi] The Effect of Application of an Idea
Jim, You are absolutely correct. Try the following logical definition/meme-map where all links are tautologies (and therefore transitive) -Spread || | | || | | vvv v Intelligent -- Friendly -- Ethical -- {Core of all religions}--{Any current religion minus irrational, unethical , stupidities} ^ ^ \^ ^ | | \ | | v v\ v v Rational--Self-Interest {Play Well With Others)--Love The meme itself is a successful implementation of Seed Friendliness (and Intelligence) if successfully implanted. - Original Message - From: Jim Bromer To: agi@v2.listbox.com Sent: Saturday, March 22, 2008 5:33 PM Subject: **SPAM** [agi] The Effect of Application of an Idea I want to quickly mention an idea I had a few years ago. New knowledge can be thought of as data stored somewhere in some kind of database, but it also has a greater potential of effect when it used intelligently. New knowledge is not just a dull object of information to be stored, because it has a potential to be far reaching when it is used. It even has the potential to change the course of ones thinking. Knowledge is varied, and the usefulness, adequacy and accuracy will not be the same for each piece of information. Some information will constitute a meaningful idea and some will only constitute a fragment of an idea. Knowledge constitutes understanding only when it is combined with more knowledge that is relevant to it and can be used effectively to integrate it into a greater sense of the subject matter. So even though a piece of knowledge may be recognized as meaningful, that meaning exists because it can be related to many other kinds of knowledge. Knowledge is not just a piece of information that lies dully on a dusty shelf, so to speak, it has the potential to dynamically relate to other kinds of knowledge. In my view of advanced AI, knowledge has to play different computational roles with other kinds of knowledge. A few years ago, I realized that different kinds of knowledge can have a varied range of effect when it is applied to other knowledge. Let's say that I want to store the information that parts x1332-b and z733-c are somehow related into a database record. If no other kind of information has been stored about those parts, or if no relations with other relevant information has been programmed into the database, then that information will exist only as a rather dull fragment of information that is unrelated to any other information. And the vast majority of people who will read this will quickly forget that parts x1332-b and z733-c are somehow related just because it holds so little meaning for them. On the other hand, if I convince you that you should be a little more open minded, you might take that thought and apply it to a wide variety of situations. The difference is that a persuasive argument to change the way you think about things in general can potentially have a greater effect of application than a piece of information concerning some trivial fact that is has little meaning to you. One of the great things about this theory of the effect of application is that it can be modeled in a simple closed artificial system and studied, but it can also subsequently be used in an extension of the system. It is in effect, a foundation of conceptual integration, and by recognizing the significance of the theory, I believe that some important new areas of research may be discovered. The theory can help advanced students imaginatively discover how ideas can play different roles with other ideas and help them to see beyond the narrow application of fundamental primitives that dominate (and, in my opinion, often disable) current research into AGI. Although the theory can be used as a fundamental primitive in simple experiments, it can also be used in the definition of extensive conceptual systems as well. It's beauty that it is not just a fundamental piece of knowledge; it is also a principle in a theory of conceptualization. Ideas have effects on other ideas. Jim Bromer -- agi | Archives | Modify Your Subscription --- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: