Re: [agi] Philosophy of General Intelligence
Matt, Suppose you write a program that inputs jokes or cartoons and outputs whether or not they are funny. Then there is an iterative process by which you can create funny jokes or cartoons. Write a program that inputs a movie and outputs a rating of 1 to 5 stars. Then you have an iterative process for creating good movies. The system first needs to parse the input and translate it into its KR. For movies - no way at this point because of technology limitations (even if we had KR format that could well express it). Jokes in NL - still a problem (decades of trouble with NL as you know - there is a good reason for that). Jokes in a formal language - that could work IF we get the KR right. There are many types of Jokes. Each type has its algorithm and the algorithms can be combined. Simple algorithm example: Comparison of 2 objects which have some of the same (or very similar) characteristics. Emphasizing the similarity (= Optional part 1.). Then applying non-identical characteristic(s) of object1 in an action taken by the object2 (pretending that the object2 also has that characteristic) and deriving a result which is in contrast with the result we would get if it was for real. (= Part 2.). If you have lots of data and a decent KR then you can query it for data to fill joke templates (+ use various modifiers for uniqueness), detect and rate jokes. Funny stuff is often based on contrast unexpected turns. Also certain creatures (like ducks) have often a better potential than others. And of course, there are also certain things in particular societies you need to avoid. If the system gets feedback joke samples, it can tweak/generate its joke templates (always considering info about the audience) and get better. Decent KR - that's the first thing. Regards, Jiri Jelinek --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] Re: AI isn't cheap
Matt, On 9/7/08, Matt Mahoney [EMAIL PROTECTED] wrote: --- On Sun, 9/7/08, Steve Richfield [EMAIL PROTECTED] wrote: 1. I believe that there is some VERY fertile but untilled ground, which if it is half as good as it looks, could yield AGI a LOT cheaper than other higher estimates. Of course if I am wrong, I would probably accept your numbers. 2. I believe that AGI will take VERY different (cheaper and more valuable) forms than do other members on this forum. Each of the above effects are worth several orders of magnitude in effort. You are just speculating. Of course. Aren't we all here on this forum? The fact is that thousands of very intelligent people have been trying to solve AI for the last 50 years, and most of them shared your optimism. Unfortunately, their positions as students and professors at various universities have forced almost all of them into politically correct paths, substantially all of which lead nowhere, for otherwise they would have succeeded long ago. The few mavericks who aren't stuck in a university (like those on this forum) all lack funding. Perhaps it would be more fruitful to estimate the cost of automating the global economy. I explained my estimate of 10^25 bits of memory, 10^26 OPS, 10^17 bits of software and 10^15 dollars. I don't understand the goal or value here? Perhaps you could explain? You really should see my Dr. Eliza demo. Perhaps you missed my comments in April. http://www.listbox.com/member/archive/303/2008/04/search/ZWxpemE/sort/time_rev/page/2/entry/5:53/20080414221142:407C652C-0A91-11DD-B3D2-6D4E66D9244B/ Apparently I did. Sorry about that. Here I have pasted in the posting with embedded contemporary comments. --- Steve Richfield [EMAIL PROTECTED] wrote: Why go to all that work?! I have attached the *populated* Knowledge.mdb file that contains the knowledge that powers the chronic illness demo of Dr. Eliza. To easily view it, just make sure that any version of MS Access is installed on your computer (it is in Access 97 format) and double-click on the file. From there, select the Tables tab, and click on whatever table interests you. I looked at your file. Would I be correct that if I described a random health problem to Dr. Eliza that it would suggest that my problem is due to one of: - Low body temperature - Fluorescent lights - Consuming fructose in the winter - Mercury poisoning from amalgam fillings and vaccines - Aluminum cookware - Hydrogenated vegetable oil - Working a night shift - Aspirin (causes macular degeneration) - Or failure to accept divine intervention? = First, my complements on your careful reading of the knowledge base. = Yes, there is a pretty good chance that you would be asked about some of these things, as various of these things seem to underlie most chronic illnesses. Is that it, or is there a complete medical database somewhere, = WYSIWYG, though this is only maybe 1% of a fully populated health database. This stuff is just there for demo. For a 1% demo, it works amazingly well. Further, I presume that people would embed generous hyperlinks into the explanations, so that Pub Med and other medical databases would be just a mouse click away. or the capability of acquiring this knowledge? = Only machine knowledge that has been carefully crafted by humans. As I have explained in a number of postings, certain key things, like how people commonly express symptoms and the carefully crafted questions needed to drill down, are NOT on any web site or medical text, so the services of an experienced expert is absolutely required. Plans of others to mine the Internet (or Wikipedia) are absolutely doomed to failure because this information is so completely lacking. No AGI would be able to compose this knowledge unless they had the real-world experience with real-world people to know how they express things. In short, many AI and AGI plans are quite obviously hopeless because they lack access to this information. Do you have a medical background, = Yes. or have you consulted with doctors in building the database? = Yes. BTW, regarding processes that use 100% of CPU in Windows. Did you try Ctrl-Alt-Del to bring up the task manager, then right click on the process and change its priority? = Not that specifically, though I did try the Windows API to do the same, and got back an error code that indicated that the most problematical task (NaturallySpeaking) had set a bit to keep other tasks from adjusting its priority. I presume that the Task Manager would have simply called the same API but probably failed to provide the return code. I expect to abandon speech I/O in the future even though it works pretty well, because no one seems to want to bet their success in overcoming their problems on the random screwups of a speech recognition program. Without speech I/O, there is no speed problem. This is apparently one of those great ideas that just can't make it in the real world. In any case,
Re: [agi] Does prior knowledge/learning cause GAs to converge too fast on sub-optimal solutions?
Hi, I am curious about the result you mention. You say that the genetic algorithm stopped search very quickly. Why? It sounds like they want to search to go longer, but can't they just tell it to go longer if they want it to? And to reduce convergence, can't they just increase the level of mutation? Do you know if they tried this, and if so, why it wasn't sufficient? Other than that, I think there are several things to try. First, it seems more natural to me to put the textbook solutions in the initial population, rather than coding them as genetic operations. Second, if they are used as operations, I'd try splitting them up further (just to reduce the bias). Disclaimer: I do not consider myself an expert, as I am still an undergraduate. --Abram On Sun, Sep 7, 2008 at 8:55 PM, Benjamin Johnston [EMAIL PROTECTED] wrote: Hi, I have a general question for those (such as Novamente) working on AGI systems that use genetic algorithms as part of their search strategy. A GA researcher recently explained to me some of his experiments in embedding prior knowledge into systems. For example, when attempting to automate the discovery of models of a mechanical system, they tried adding some textbook models to the set of genetic operators. The results weren't good – the prior knowledge worked too well, causing the GA to converge too fast onto the prior knowledge… so fast that there wasn't time for the GA to build up sufficient diversity and quality in other solutions that might have helped get out of the local maxima. The message seemed to be that prior knowledge is too powerful – it can 'blind' a search – and that if you must use it, you'd have to very very aggressively artificially deflate the fitness of instances that use prior knowledge (and this is tricky to get right). This struck me as relevant to GA-based AGIs that continually build on and improve a knowledge-base. Once an AGI learns very simple initial models of the world, if it then tries to evolve deeper knowledge about more difficult problems (but, in the context of its prior learning), then its initial models may prove to be too good: forcing the GA to converge on poor local maxima that represent only minor variations on the initial models it learnt in its earliest days. Does this issue actually crop up in GA-based AGI work? If so, how did you get around it? If not, would you have any comments about what makes AGI special so that this doesn't happen? -Ben agi | Archives | Modify Your Subscription --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
RE: Language modeling (was Re: [agi] draft for comment)
From: Matt Mahoney [mailto:[EMAIL PROTECTED] --- On Sun, 9/7/08, John G. Rose [EMAIL PROTECTED] wrote: From: John G. Rose [EMAIL PROTECTED] Subject: RE: Language modeling (was Re: [agi] draft for comment) To: agi@v2.listbox.com Date: Sunday, September 7, 2008, 9:15 AM From: Matt Mahoney [mailto:[EMAIL PROTECTED] --- On Sat, 9/6/08, John G. Rose [EMAIL PROTECTED] wrote: Compression in itself has the overriding goal of reducing storage bits. Not the way I use it. The goal is to predict what the environment will do next. Lossless compression is a way of measuring how well we are doing. Predicting the environment in order to determine which data to pack where, thus achieving higher compression ratio. Or compression as an integral part of prediction? Some types of prediction are inherently compressed I suppose. Predicting the environment to maximize reward. Hutter proved that universal intelligence is a compression problem. The optimal behavior of an AIXI agent is to guess the shortest program consistent with observation so far. That's algorithmic compression. Oh I see. Guessing shortest program = compression. OK right. But yeah like Pei said the word compression is misleading. It implies a reduction where you are actually increasing understanding :) John --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
[agi] Will AGI Be Stillborn?
From the article: A team of biologists and chemists [lab led by Jack Szostak, a molecular biologist at Harvard Medical School] is closing in on bringing non-living matter to life. It's not as Frankensteinian as it sounds. Instead, a lab led by Jack Szostak, a molecular biologist at Harvard Medical School, is building simple cell models that can almost be called life. http://blog.wired.com/wiredscience/2008/09/biologists-on-t.html There's a video entitled A Protocell Forming from Fatty Acids. It's fascinating and, at the same time, a bit scary. Paper co-authored by Szostak published this month: Thermostability of model protocell membranes http://www.pnas.org/content/early/2008/09/02/0805086105.full.pdf+html --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] draft paper: a hybrid logic for AGI
A somewhat revised version of my paper is at: http://www.geocities.com/genericai/AGI-ch4-logic-9Sep2008.pdf (sorry it is now a book chapter and the bookmarks are lost when extracting) On Tue, Sep 2, 2008 at 7:05 PM, Pei Wang [EMAIL PROTECTED] wrote: I intend to use NARS confidence in a way compatible with probability... I'm pretty sure it won't, as I argued in several publications, such as http://nars.wang.googlepages.com/wang.confidence.pdf and the book. I understood your argument about defining the confidence c, and agree with it. But I don't see why c cannot be used together with f (as *traditional* probability). In summary, I don't think it is a good idea to mix B, P, and Z. As Ben said, the key is semantics, that is, what is measured by your truth values. I prefer a unified treatment than a hybrid, because the former is semantically consistent, while the later isn't. My logic actually does *not* mix B, P, and Z. They are kept orthogonal, and so the semantics can be very simple. Your approach mixes fuzziness with probability which can result in ambiguity in some everyday examples: eg, John tries to find a 0.9 pretty girl (degree) vs Mary is 0.9 likely to be pretty (probability). The difference is real, but subtle, and I agree that you can mix them but you must always acknowledge that the measure is mixed. Maybe you've mistaken what I'm trying to do, 'cause my theory should not be semantically confusing... YKY --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] draft paper: a hybrid logic for AGI
On Tue, Sep 2, 2008 at 12:05 PM, Ben Goertzel [EMAIL PROTECTED] wrote: but in a PLN approach this could be avoided by looking at IntensionalInheritance B A rather than extensional inheritance.. The question is how do you know when to apply the intensional inheritance, instead of the extensional one. It seems to me that using the probabilistic interpretation of fuzziness would force you to use sum-product calculus... YKY --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] draft paper: a hybrid logic for AGI
Sorry I don't have the time to type a detailed reply, but for your second point, see the example in http://www.cogsci.indiana.edu/pub/wang.fuzziness.ps , page 9, 4th paragraph: If these two types of uncertainty [randomness and fuzziness] are different, why bother to treat them in an uniform way? The basic reason is: in many practical problems, they are involved with each other. Smets stressed the importance of this issue, and provided some examples, in which randomness and fuzziness are encountered in the same sentence ([20]). It is also true for inferences. Let's take medical diagnosis as an example. When a doctor want to determine whether a patient A is suffering from disease D, (at least) two types of information need to be taken into account: (1) whether A has D's symptoms, and (2) whether D is a common illness. Here (1) is evaluated by comparing A's symptoms with D's typical symptoms, so the result is usually fuzzy, and (2) is determined by previous statistics. After the total certainty of A is suffering from D is evaluated, it should be combined with the certainty of T is a proper treatment to D (which is usually a statistic statement, too) to get the doctor's degree of belief for T should be applied to A. In such a situation (which is the usual case, rather than an exception), even if randomness and fuzziness can be distinguished in the premises, they are mixed in the middle and final conclusions. Pei On Mon, Sep 8, 2008 at 3:55 PM, YKY (Yan King Yin) [EMAIL PROTECTED] wrote: A somewhat revised version of my paper is at: http://www.geocities.com/genericai/AGI-ch4-logic-9Sep2008.pdf (sorry it is now a book chapter and the bookmarks are lost when extracting) On Tue, Sep 2, 2008 at 7:05 PM, Pei Wang [EMAIL PROTECTED] wrote: I intend to use NARS confidence in a way compatible with probability... I'm pretty sure it won't, as I argued in several publications, such as http://nars.wang.googlepages.com/wang.confidence.pdf and the book. I understood your argument about defining the confidence c, and agree with it. But I don't see why c cannot be used together with f (as *traditional* probability). In summary, I don't think it is a good idea to mix B, P, and Z. As Ben said, the key is semantics, that is, what is measured by your truth values. I prefer a unified treatment than a hybrid, because the former is semantically consistent, while the later isn't. My logic actually does *not* mix B, P, and Z. They are kept orthogonal, and so the semantics can be very simple. Your approach mixes fuzziness with probability which can result in ambiguity in some everyday examples: eg, John tries to find a 0.9 pretty girl (degree) vs Mary is 0.9 likely to be pretty (probability). The difference is real, but subtle, and I agree that you can mix them but you must always acknowledge that the measure is mixed. Maybe you've mistaken what I'm trying to do, 'cause my theory should not be semantically confusing... YKY --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] draft paper: a hybrid logic for AGI
On Tue, Sep 9, 2008 at 4:27 AM, Pei Wang [EMAIL PROTECTED] wrote: Sorry I don't have the time to type a detailed reply, but for your second point, see the example in http://www.cogsci.indiana.edu/pub/wang.fuzziness.ps , page 9, 4th paragraph: If these two types of uncertainty [randomness and fuzziness] are different, why bother to treat them in an uniform way? The basic reason is: in many practical problems, they are involved with each other. Smets stressed the importance of this issue, and provided some examples, in which randomness and fuzziness are encountered in the same sentence ([20]). It is also true for inferences. Let's take medical diagnosis as an example. When a doctor want to determine whether a patient A is suffering from disease D, (at least) two types of information need to be taken into account: (1) whether A has D's symptoms, and (2) whether D is a common illness. Here (1) is evaluated by comparing A's symptoms with D's typical symptoms, so the result is usually fuzzy, and (2) is determined by previous statistics. After the total certainty of A is suffering from D is evaluated, it should be combined with the certainty of T is a proper treatment to D (which is usually a statistic statement, too) to get the doctor's degree of belief for T should be applied to A. In such a situation (which is the usual case, rather than an exception), even if randomness and fuzziness can be distinguished in the premises, they are mixed in the middle and final conclusions. Thanks, that's a good point that I haven't thought of. For example I have a _slight_ knee pain (fuzzy, z = 0.6) knee pain - rheumatoid arthritis (p = 0.3) (excuse me for making up numbers) Then my system would convert knee pain (z = 0.6) to knee pain = true (binary) and conclude rheumatoid arthritis (p = 0.3) So there is some loss of information, but I feel this is OK. Many commonsense reasoning steps are lossy. We're not trying to build doctors here. A commonsense AGI can control a medical expert system to achieve professional levels. The point is, I can always keep P and Z orthogonal. YKY --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] Will AGI Be Stillborn?
I've reflected that superintelligence could emerge through genetic or pharmaceutical options before cybernetic ones, maybe by necessity. I am really rooting for cybernetic enlightenment to guide our use of the other two, though. On 9/8/08, Brad Paulsen [EMAIL PROTECTED] wrote: From the article: A team of biologists and chemists [lab led by Jack Szostak, a molecular biologist at Harvard Medical School] is closing in on bringing non-living matter to life. It's not as Frankensteinian as it sounds. Instead, a lab led by Jack Szostak, a molecular biologist at Harvard Medical School, is building simple cell models that can almost be called life. http://blog.wired.com/wiredscience/2008/09/biologists-on-t.html There's a video entitled A Protocell Forming from Fatty Acids. It's fascinating and, at the same time, a bit scary. Paper co-authored by Szostak published this month: Thermostability of model protocell membranes http://www.pnas.org/content/early/2008/09/02/0805086105.full.pdf+html --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] Does prior knowledge/learning cause GAs to converge too fast on sub-optimal solutions?
You can implement a new workaround to bootstrap your organisms past each local maximum, like catalyzing the transition from water to land over and over. I find this leads to cheats that narrow the search in unpredictable ways, though. This problem comes up again and again. Maybe some kind of drift in the parameters or fitness function would destabilize deeply-converged positions. I've thought before how useful it would be to have an AI tuning my GA. _o On 9/8/08, Benjamin Johnston [EMAIL PROTECTED] wrote: I am curious about the result you mention. You say that the genetic algorithm stopped search very quickly. Why? It sounds like they want to search to go longer, but can't they just tell it to go longer if they want it to? They found that the system converged too quickly. The initial knowledge quickly dominated the population, and then successive generations showed little improvement. And to reduce convergence, can't they just increase the level of mutation? Do you know if they tried this, and if so, why it wasn't sufficient? The quality of the solutions found using prior knowledge was such that any random mutations was almost always inferior. As I understood it, to get out of the local maxima that prior knowledge gets a GA stuck in, you really need some reasonable quality solutions so that larger structures of a good solution can be introduced via cross-over. Any given random mutation was usually detrimental - real progress depended on a child being able to combine complex substructures from two different parents. Other than that, I think there are several things to try. First, it seems more natural to me to put the textbook solutions in the initial population, rather than coding them as genetic operations. Second, if they are used as operations, I'd try splitting them up further (just to reduce the bias). Yes, those are good points - I have been wondering about that, but I didn't have the chance to ask those questions. Presumably one problem is that if you just put prior knowledge in the initial population, unmatched to the system parameters, then the textbook models would be unreasonably bad; they would quickly be eliminated and there would be little chance for them to be reintroduced later into the population. One solution to this might then be to have a fixed 'immortal' population of textbook models that can be crossed with the rest of the population at any time. Another possibility could be to use island-GA, with prior knowledge 'banned' from some of the islands. Anyway, I'm sure there must be lots of different ways that sound like they might solve the problem. But, which (or whether any) ones actually work in practice is another matter. And that's why I'm curious to know whether AGI researchers have encountered this problem, and what they have done about it... -Ben --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com