RE: [agi] Do the inference rules.. P.S.Edward, Thankyou for you putting your POV very clearly - and the hope perhaps of most AGI-ers still.
I think you are dead wrong - and it's the most expensive professional mistake of any AGI'er's life. I would suggest thinking a great deal further about all this. IOW you are wrong in thinking that thought about the world (or domains thereof) can take place without sensory knowledge/ images of the world (or those domains) - and without the or any mind simultaneously making sense of its symbols. Words and other symbols are just "names" for objects and aspects thereof. (Or labels on boxes). If you haven't seen or sensed the actual object behind the name, (or seen inside the box), those symbols are "senseless". [Slogan: The name is not the object. The name or word tells you nothing about the object it refers to.] You haven't seen (I presume) "Vladimir Nesov" and "Mike Tintner" - so your capacity to infer further information (especially physical) about the objects/people behind the names is extremely limited. (Nevertheless your mind will have already posited tentative if vague pictures for both!). You will probably have inferred a great deal about our personalities from our posts , but all that will be based on your sensory experience of people using words and expressing ideas in the flesh - experience, for example, that people who say "you are dead wrong" are usually fairly aggressive types. A set of symbols for which you have no sensory image of the named object at all is totally senseless. Tell me about "Vladimir Imrov." Nothing, right? But did you know that he has a "grundchen"? And that belongs to the class of "oblomovs"? And that's a very interesting class because it encompasses both "imries" and "collusors" as well as "grundchens". Grundchens are partly imries and partly collusors. If I had to calculate the possibility of an imrie being a collusor, though, I guess it would be roughly 40%, provisionally - as long as it doesn't have an "exxen".... Now you know what it's like to be NARS and every other AGI computer. In the final analysis, computers at present are just fancy label-readers and comparers. They don't know wtf they're doing. [Note BTW how your mind started to go blank while reading the last para]. Come to your senses. And find a way to give computers some (i.e. for grounded thought).. P.S. To be an AGI and solve higher adaptive problems, you have to be able to come up with NEW information about the world (or a scene) - in order to find new kinds of solution. The only way that "senseless" computers like NARS can do that - as far as I understand - is by by looking at their hierarchical charts and networks of "oblomovs", "grundchens," "imries,"collusors" and "exxens", and noticing new connections - noticing, say, than an "exxen" also connects up with and has a certain degree of "grundchen-ness" (whatever that means). Humans and animals however infer new information to a massive extent by LOOKING at the world (or that scene) - even if only in our mind's eye - and SEEING/sensing new approaches to problems - (although we are often under the illusion that it's all happening on a purely symbolic level). I've tried to get others here to give me some examples of how symbolic analogies can be fruitful without success. (I'm not saying they can't be at all, just very limited). You, Edward, might care to engage with that challenge. I think it's an important one worth exploring. ----- Original Message ----- From: Edward W. Porter To: [email protected] Sent: Thursday, October 11, 2007 1:13 PM Subject: RE: [agi] Do the inference rules.. P.S. Validimir and Mike, For humans, much of our experience is grounded on sensory information, and thus much of our understanding is based on experiences and analogies derived largely from the physical world. So Mike you are right that for us humans, much of our thinking is based on recasting of experiences of the physical world. But just because experience of the physical world is at the center of much of human thinking, does not mean it must be at the center of all possible AGI thinking -- any more than the fact that for millions of years the earth and the view from it was at the center of our thinking and that of our ancestors means the earth and the view from it must forever be at the center of the thinking of all intelligences throughout the universe. In fact, one can argue that for us humans, one of our most important sources of grounding - emotion -- is not really about the physical world (at least directly), but rather about our own internal state. Furthermore, multiple AGI projects, including Novamente and Joshua Blue are trying to ground their systems from experience in virtual words. Yes those virtual worlds try to simulate physical reality, but the fact remains that much of the grounding is coming from bits and bytes, and not from anything more physical. Take Doug Lenat's AM and create a much more powerful AGI equivalent of it, one with much more powerful learning algorithms (such as those in Novamente), running on the equivalent of a current 128K processor BlueGene L with 16TBytes of RAM, but with a cross sectional bandwidth roughly 500 times that of the current BlueGene L (the type of hardware that could be profitably sold for well under 1 million dollars in 7 years if there were are thriving market for making hardware to support AGI). Assume the system creates programs, mathematical structures, and transformations, etc. and in its own memory. It starts out learning like a little kid, constantly performing little experiments, except the experiments -- instead of being things like banging spoons against a glass -- would be running programs that create data structures and then observing what is created (it would have built in primitives for observing its own workspace), changing the program and observing the change, etc. Assume it receives no input from the physical world, but that it has goals and a reward system related to learning about programming, finding important mathematical and programming generalities, finding compact representations and transformation, creating and finding patterns in complexity, and things like that. Over time such a system would develop its own type of grounding, one derived from years of experience -- and from billions of trillions of machine opps -- in programming and math space. Thus, I think you are both right. Mike is right that for humans, sensory experience is a vital part of much of our ability to understand, even of our ability to understand things that might seem totally abstract. But Validmir is right for believing that it should be possible to build an AGI that was well grounded in its own domain, without any knowledge of the physical world (other than as the manifesting of bits and bytes). Edward W. Porter Porter & Associates 24 String Bridge S12 Exeter, NH 03833 (617) 494-1722 Fax (617) 494-1822 [EMAIL PROTECTED] -----Original Message----- From: Mike Tintner [mailto:[EMAIL PROTECTED] Sent: Wednesday, October 10, 2007 11:10 PM To: [email protected] Subject: Re: [agi] Do the inference rules.. P.S. Vladimir, I'm not trying to be difficult or critical but I literally can't understand what you're saying, because you haven't given any example of a problem, where "knowledge of concepts' relations and implications" somehow supersedes or is independent of physical casting/ recasting. Your analogy though of what I might be saying about maths (or other symbols) is wrong. Numbers and arithmetic are based on and derive from physical objects and our ability to add and subtract objects etc . Geometry is obviously based on an analysis of physical objects and shapes. They are totally physically object-based and can only be understood as such. To point this out is not at all the same as suggesting that their figures are composed of ink. I am talking about what their figures (and other symbols like language) refer to, not what they are composed of. (Even a mathematical concept BTW like "infinity" only became acceptable in maths about the time of the printing press - when it became possible physically/ realistically for the first time to imagine objects being produced ad infinitum). And I would suggest that our ability to perceive the kinds of concept relations you may be thinking of is very much physically based and "digital" - IOW based on pointing with our digits to different objects in a scene (even if only in our mind's eye) - to explain, for example, by pointing to how "this moves that" and classify by pointing out that "he is the parent of her" etc. Your ink analogy BTW is also, very much I suggest, a physically based and technically fascinating operation. How do you think you arrived at it - other than physically and spatially? Do you think you could arrive at such an analogy simply by comparing sets of symbolic properties of scientific problems and their physical recasting , on the one hand, and symbolic properties of numbers and ink on the other? (Such an analogy is where the Fauconnier-style analysis of "conceptual blending" of "mental spaces" comes into its own. How the brain achieves an analogy as complex as your ink one is still something quite awesome and problematic, even with Fauconnier's help - and still way beyond computers, I suggest). Vladimirwrote: >> Vladimir, >> >> No I'm sure the problem-solving isn't all down to recasting in terms >> of physical models. But can you think of a scientific problem area, >> where such recasting isn't involved? > > I just tried to provide my reason for considering it a mirage: even if > problem-solving doesn't involve physical reasoning, it can be > introspectively recast as sequence of spatial representations. > Domain-specific concepts involved in problem-solving can easily be > placed on spatially arranged schemata, but reasoning is correctly > carried out because of knowledge of these concepts' relations and > implications, not simply because of spatial setting in which they are > arranged. It's equivalent to calling mathematics ink-reasoning because > it's historically performed with help of remarks made by ink on paper > and any result can be written down by ink on paper. > >> >> (Very tangentially, what comes to my mind is chess. I'm confident >> that >> human >> problemsolving here - and the ability to search through only scores as >> opposed to billions of chessboard scenarios to arrive at moves - >> depends >> on physical models, and is an ahem graphic illustration of the very >> different ways in which current computers and a true general intelligence >> think). >> >> Vladimir: These 'recastings' of problems are essentially inference >> steps, where >> > each step is evident and is performed by trained expert's >> > intuition. Sequence of such simple steps can constitute complex >> > inference which leads to solution of complex problem. This >> > recasting isn't necessarily related to physical common sense, even >> > though each intermediate representation can be represented as >> > spatially-temporal construction by virtue of being representable by >> > frame graphs evolving over time, which does not reflect the rules >> > of this evolution (which are the essence of inference which is >> > being performed). >> > >> > On 10/11/07, Mike Tintner <[EMAIL PROTECTED]> wrote: >> >> Just to underline my point about the common sense foundations of >> >> logic and general intelligence - I came across this from : >> >> Education & Learning to >> >> Think by Lauren B Resnick - (and a section entitled "General >> >> Reasoning - >> >> Improving Intelligence). >> >> >> >> "Recent research in science problem solving shows that experts do >> >> not respond to problems as they are presented - writing equations >> >> for every relationship described and then using routine procedures >> >> for manipulating >> >> equations.Instead they reinterpret the problems, recasting them in >> >> terms >> >> of >> >> general scientific principles until the solutions become almost >> >> self-evident." >> >> >> >> He points out that the same principles apply to virtually all >> >> subjects >> >> in >> >> the curriculum. I would suggest that those experts are recasting >> >> problems >> >> principally in terms of physical common sense models. NARS, it seems >> >> to >> >> me, >> >> "responds to problems as they are presented." >> >> >> >> >> >> ----- >> >> 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/?&; >> >> >> > >> > >> > -- >> > Vladimir Nesov mailto:[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/?&; >> > >> > >> > >> > -- >> > No virus found in this incoming message. >> > Checked by AVG Free Edition. >> > Version: 7.5.488 / Virus Database: 269.14.6/1060 - Release Date: >> > 09/10/2007 16:43 >> > >> > >> >> >> ----- >> 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/?&; >> > > > -- > Vladimir Nesov mailto:[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/?&; > > > > -- > No virus found in this incoming message. > Checked by AVG Free Edition. > Version: 7.5.488 / Virus Database: 269.14.6/1060 - Release Date: > 09/10/2007 16:43 > > ----- 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/?&; ------------------------------------------------------------------------------ No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.488 / Virus Database: 269.14.7/1062 - Release Date: 10/10/2007 17:11 ----- 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=8660244&id_secret=52296606-900881
