RE: [agi] WordNet and NARS
hi I'm also extremely interested in biological applications, but maybe your choice is biased by your (emotional) preference for science and is actually not optimal / rational. I'm sure it's rational; of course I'm not sure it's *optimal*! Business is not a very exact science, and one is always making choices based on intuitive assessments of a large number of complexly interrelated factors. Optimality is really never an issue. A key point is that I was able to think of a lot of concrete ways to use AI to make marketable products in the biology domain. This is because I'm a scientist and have a good intuition for such things. So the basis of our company Biomind LLC is not just our Novamente AI system but -- just as importantly -- a host of other scientific ideas I had that bridge the gap between Novamente's general capabilities and the specific problems involved in the bio domain. Having knowledge of and intuition for the domain of application is probably THE MOST IMPORTANT THING in doing practical AI applications --- be they commercial or academic, be they proto-AGI or pure-narrow-AI. By looking at some employment statistics, I guesstimate the most profitable AI applications should be retail and food industries, and slowly work its way up more complex areas (eg health care such as robotic surgery or medical expert systems etc), and then even more complex ones such as those requiring natural language understanding, and other cognitive abilities. But this kind of analysis is so extremely coarse that I'm not sure it's useful at all. The pragmatics of selling products into these different industries needs to be taken into account. I really don't know a good way to sell AI in the retail and food industries, for example; whereas I do know how to sell AI-driven bioinformatics tools into the biopharma market. Sure, you could try to sell AI-based datamining to Wal-mart and the like. But I'm not sure this is really a better business proposition than selling AI-based tools to biopharma? One advantage of selling to biopharma is that the people involved are scientists and hence are generally more open to radical ideas than average business-types would be. If anyone on this list has good connections with the datamining-related people at Walmart or other major retail chains, I'd be happy to meet with them and pitch them proto-AGI-based-datamining. I'd be happy to take a temporary (paid) contract to analyze some data for them, just to show them what our software can do on their data ;-) Unless you think AGI does not need this bottom-up approach... I think there are many paths to AGI, not just one And I think that NO business application is going to take us all the way to AGI -- doing business apps can help us develop various aspects of our proto-AGI systems, but ultimately you've got to teach a baby mind and that is a research project not a business project, due to the very many uncertainties involved. We are doing business apps like Biomind because they're fascinating valuable in themselves (extending life, curing diseases, etc.), because they help us develop and test various *parts* of our AI systems, and because they pay our salaries ;-) But it's a mistake to try to map the business-app goals too closely onto the AGI goals -- although it's great when the two harmonize, as is somewhat the case with our Biomind work right now... ben --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] WordNet and NARS
From: Ben Goertzel [EMAIL PROTECTED] So far our work in this area has been more in the vein of narrow AI using a half-completed wannabe-AGI system, but I'm curious to see how the molecular biology software applications make use of the AGI capabilities of Novamente when/if they finally become real... Quite frankly, our motivation for working in this area has largely been pragmatic. We couldn't seem to get funding for pure AI research so we had to choose one or more application areas. I'm very interested in life extension research and in understanding the molecular basis of life generally, so applying our AI tech in these domains seemed a natural. I'm also extremely interested in biological applications, but maybe your choice is biased by your (emotional) preference for science and is actually not optimal / rational. By looking at some employment statistics, I guesstimate the most profitable AI applications should be retail and food industries, and slowly work its way up more complex areas (eg health care such as robotic surgery or medical expert systems etc), and then even more complex ones such as those requiring natural language understanding, and other cognitive abilities. Unless you think AGI does not need this bottom-up approach... but who knows if we have not underestimated the complexity of AGI? YKY Find what you are looking for with the Lycos Yellow Pages http://r.lycos.com/r/yp_emailfooter/http://yellowpages.lycos.com/default.asp?SRC=lycos10 --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] WordNet and NARS
Ben said: However, we need to remember that the knowledge in an AGI should be *experientially grounded*. . . . but it needs to turn this knowledge into knowledge by crosslinking a decent fraction of it with perceptual and procedural patterns . . . Can a color-blind man understand yellow? Perhaps not in the same way a normal person can. But he could easily know more about yellow than many. Its wavelength, its history of use in fine arts, its psychological impact, and so on. He could even effectively use yellow in graphics, perhaps with a tool to identify yellow with a special texture. So, even though the color-blind (or an AI entity) never actually sees yellow, he can experience yellow by way of external knowledge. Perhaps the limit to this grounding by knowledge phenomenon is very high. Maybe as Ben says, the grounding can be by procedural patterns. WordNet type knowledge (implemented in a system such as NARS) could be a link to human knowledge. A yellow filter in the sights of a target rifle makes the target-sight image more distinct in low light. While I have never experienced it myself, the book in which I found this information is a standard reference for Olympic caliber competitors. So as a NARS based intelligence, I give this belief f and c values of .99 :-) Kevin Copple To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] WordNet and NARS
I agree that not all knowledge in a mind needs to be grounded. However, I think that a mind needs to have a LOT of grounded knowledge, in order to learn to reason usefully. It can then transfer some of the thinking-ability (and some of the concrete relationships) learned on the grounded domains, to help it think about its ungrounded knowledge... ben g -Original Message-From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]On Behalf Of [EMAIL PROTECTED]Sent: Wednesday, February 04, 2004 3:23 AMTo: [EMAIL PROTECTED]Subject: RE: [agi] WordNet and NARS Ben said: However, we need to remember that the knowledge in an AGI should be *experientially grounded*. . . . but it needs to turn this "knowledge" into knowledge by crosslinking a decent fraction of it with perceptual and procedural patterns . . . Can a color-blind man understand yellow? Perhaps not in the same way a normal person can. But he could easily know more about yellow than many. Its wavelength, its history of use in fine arts, its psychological impact, and so on. He could even effectively use yellow in graphics, perhaps with a tool to identify yellow with a special texture. So, even though the color-blind (or an AI entity) never actually sees yellow, he can experience yellow by way of external knowledge. Perhaps the limit to this grounding by knowledge phenomenon is very high. Maybe as Ben says, the grounding can be by procedural patterns. WordNet type knowledge (implemented in a system such as NARS) could be a link to human knowledge. A yellow filter in the sights of a target rifle makes the target-sight image more distinct in low light. While I have never experienced it myself, the book in which I found this information is a standard reference for Olympic caliber competitors. So as a NARS based intelligence, I give this belief f and c values of .99 :-) Kevin Copple To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED] To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] WordNet and NARS
Philip, I think it's important for a mind to master SOME domain (preferably more than one), because advanced and highly effective cognitive schemata are only going to be learned in domains that have been mastered. These cognitive schemata can then be applied in other domains as well, which are understood only to a lesser degree of mastery. And, as you say, in order for the AI to master some domain, it needs a lot of grounded knowledge in that domain. So, I am skeptical that an AI can really think effectively in ANY domain unless it has done a lot of learning based on grounded knowledge in SOME domain first; because I think advanced cognitive schemata will evolve only through learning based on grounded knowledge... -- Ben So the way you describe things seems to fit the domain where an AGI is trying to build mastery but I'm not convinced that the AGI absolutely needs a high level of grounded knowledge in areas where it is not building mastery. But in areas where the AGI is not building or better still has not achieved mastery it should exercise humility and caution and not make any rash decisions that could affect others - because it really doesn't know how sensible its inherited knowledge is. This seems to me to be an area where ethics intersects with mind dvelopment and the use of mind. Cheers, Philip --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED] --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] WordNet and NARS
Hi Ben, So, I am skeptical that an AI can really think effectively in ANY domain unless it has done a lot of learning based on grounded knowledge in SOME domain first; because I think advanced cognitive schemata will evolve only through learning based on grounded knowledge... OK. I think we're getting close to agreement on most of this except what could be the key starting point. My intuition is that, if an AGI is to avoid (an admittedly accelerated) recapitulation of 3500 billion year evolution of functioning mind, it will have to start thinking *first* in one domain using inherited rules of thumb for interpreting data (and it might help to download some initial ungrounded data that otherwise would have had to be accumulated through exposure to its surroundings). Once the infant AGI has some competence using these implanted rules of thumb it can then go through the job of building it's own grounded rules of thumb for data intepretation and substituting them for the rules of thumb provided at the outset by its creators/trainers. So my guess is that the fastest (and still effective) path to learning would be: - *first* a partially grounded experience - *then* a fully grounded mastery - then a mixed learning strategy of grounded and non-grounded as need and oportunity dictates Cheers, Philip --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] WordNet and NARS
So my guess is that the fastest (and still effective) path to learning would be: - *first* a partially grounded experience - *then* a fully grounded mastery - then a mixed learning strategy of grounded and non-grounded as need and oportunity dictates Cheers, Philip Well, this appears to be the order we're going to do for the Novamente project -- in spite of my feeling that this isn't ideal -- simply due to the way the project is developing via commercial applications of the half-completed system. And, it seems likely that the initial partially grounded experience will largely be in the domain of molecular biology... at least, that's a lot of what our Novamente code is thinking about these days... -- Ben G --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] WordNet and NARS
Hi Ben, Well, this appears to be the order we're going to do for the Novamente project -- in spite of my feeling that this isn't ideal -- simply due to the way the project is developing via commercial applications of the half-completed system. And, it seems likely that the initial partially grounded experience will largely be in the domain of molecular biology... at least, that's a lot of what our Novamente code is thinking about these days... The order might be the same but I don't think the initial content will be right - unless you intend to that a conscious Novababy is born into a molecular biology world/sandbox! What were imagining the Novababy's firs simulated or real world would be? A world with a blue square and a sim-self with certain senses and actuators? Or whatever. Then that is the world I think you'll need to help the Novababy understand bu giving it ready-made rules of thumb for interpreting the data generated in that precise world. I'd be inclined to move on to a molecular biology world a little later in Novababy's life! :) Anyway - you can test my conjectures very easily with a bit of experimentation. Cheers, Philip --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] WordNet and NARS
From: Ben Goertzel [EMAIL PROTECTED] Well, this appears to be the order we're going to do for the Novamente project -- in spite of my feeling that this isn't ideal -- simply due to the way the project is developing via commercial applications of the half-completed system. And, it seems likely that the initial partially grounded experience will largely be in the domain of molecular biology... at least, that's a lot of what our Novamente code is thinking about these days... Hi Ben I'm very interested in applying automation to experimental molecular biology, especially neurobiology. I think it will help neuroscience a lot if complex experiments can be done automatically by AIs, but I'm not sure about letting an AGI reason about molecular biology in an abstract way. Which are you planning on? I'm also curious why you picked this area. YKY Get advanced SPAM filtering on Webmail or POP Mail ... Get Lycos Mail! http://login.mail.lycos.com/r/referral?aid=27005 --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] WordNet and NARS
Well, this appears to be the order we're going to do for the Novamente project -- in spite of my feeling that this isn't ideal -- simply due to the way the project is developing via commercial applications of the half-completed system. And, it seems likely that the initial partially grounded experience will largely be in the domain of molecular biology... at least, that's a lot of what our Novamente code is thinking about these days... The order might be the same but I don't think the initial content will be right - unless you intend to that a conscious Novababy is born into a molecular biology world/sandbox! That may well be the case... a robotized bio lab as an AGI playroom... we'll see! What were imagining the Novababy's firs simulated or real world would be? A world with a blue square and a sim-self with certain senses and actuators? Or whatever. Then that is the world I think you'll need to help the Novababy understand bu giving it ready-made rules of thumb for interpreting the data generated in that precise world. I think that in an environment in which the system has decent sensors and actuators, no pre-specified rules of thumb will be needed (though some perceptual preprocessing routines will be needed, just as the human visual and acoustic cortex supply ...). Pre-specified rules are useful for domains where the system's ability to perceive and act are more limited. I'd be inclined to move on to a molecular biology world a little later in Novababy's life! Well, we're already applying the incomplete AI system to molecular biology in more limited, narrow-AI-ish ways, that was my point... ben --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
RE: [agi] WordNet and NARS
Hi, WordNet is an interesting resource; we have fed it into Novamente and reasoned on it using PTL. Actually we've combined WordNet with some statistical word relationships derived from text-analysis. One runs into some memory issues on a 32-bit machine, mostly due to the bulk of the statistical relationships. Note that the SUMO ontology is crosslinked with WordNet as well. http://ontology.teknowledge.com/ This is interesting stuff, and if used properly can be 1) valuable for applications 2) good for testing certain aspects of one's cognition algorithms However, we need to remember that the knowledge in an AGI should be *experientially grounded*. A system will never achieve true intelligence with a head full of "knowledge" consisting of tokens loaded from DB's, that don't refer to any observed patterns in its experience. So it's Ok if one's AGI has a bunch of "knowledge" from WordNet, SUMO and other DB's in its mind ... but it needs to turn this "knowledge" into knowledge by crosslinking a decent fraction of it with perceptual and procedural patterns... which cannot be obtained from databases (at least not any databases currently available!!) -- Ben G -Original Message-From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]On Behalf Of [EMAIL PROTECTED]Sent: Tuesday, February 03, 2004 2:48 AMTo: [EMAIL PROTECTED]Subject: [agi] WordNet and NARS A while back, I took the WordNet database and parsed it into a relational database so that I could access it with VB. My purpose was to use it a dictionary resource for chatterbots. Then I found it could be used for other interesting things that a conventional paper dictionary cannot do very well. For example, what are the types of citrus fruit? Grouping words according to synonym sets (synsets) seems an effective means of organization, in that synsets are linked to more specific and more general sets. The synsets also include a brief definition or gloss. This type of organization gets around the problem of a word having multiple meanings; its just listed in multiple synsets. I include a couple examples of hypernym chains WordNet can produce as a postscript. It occurred to me that WordNet could be used as an ontology in which various types of information could be stored and accessed. Things as varied as concepts or individuals have a place. My hunch was that when data are organized and accessible, they could be used for a range of purposes. Another issue I have considered is how best to handle meta-data about data such as poems, books, images, and so on. At EllaZ Systems we refer to these types of data as Convuns (conversational units). Convuns tend to have a lot of properties in common, such as creator, date, type, summary, etc. When a conversation is about the Moon, for example, a number of different Convuns and types of Convuns may make appropriate fuel for conversation and interaction. In taking a closer look at NARS, it seems it could be used in a way similar to WordNet for categorizing words, concepts, and instances of information. Of course, NARS has the ability to do much more than merely categorize and store information. It should be straightforward to move the 70,000 or so synsets in WordNet into a NARS system. Perhaps this could serve as an initial grounding of a new NARS entity. For instances of information, Project Gutenberg contains thousands of public domain texts, many photos are available from public sources, and so on. Perhaps meta-data (in the form of NARS statements) about Convuns could ground them enough that a NARS (or other system) could think about them and look for patterns and understanding. Meta-data certainly helps me understand and enjoy information more! Accessible, organized information would be useful to both humans and emerging AI. Its easy to envision NARS being a big improvement over other cataloging methods, while being a part of AI development. There is certainly an appeal to the merging of data and intelligence, where the two become one. Kevin Copple P.S. A couple hypernym chains of pony are: Sense 1 (pony) A range horse of the western United States. . . . is a type of: horse, Equus caballus . . . is a type of: equine, equid . . . is a type of: odd-toed ungulate, perissodactyl, perissodactyl mammal . . . is a type of: ungulate, hoofed mammal . . . is a type of: placental, placental mammal, eutherian, eutherian mammal . . . is a type of: mammal . . . is a type of: vertebrate, craniate . . . is a type of: chordate . . . is a type of: animal, animate being, beast, brute, creature, fauna . . . is a type of: life form, organism, being, living thing . . . is a type of: entity, something For another sense of pony in another synset: