AW: AW: [agi] Re: Defining AGI
The process of outwardly expressing meaning may be fundamental to any social intelligence but the process itself needs not much intelligence. Every email program can receive meaning, store meaning and it can express it outwardly in order to send it to another computer. It even can do it without loss of any information. Regarding this point, it even outperforms humans already who have no conscious access to the full meaning (information) in their brains. The only thing which needs much intelligence from the nowadays point of view is the learning of the process of outwardly expressing meaning, i.e. the learning of language. The understanding of language itself is simple. To show that intelligence is separated from language understanding I have already given the example that a person could have spoken with Einstein but needed not to have the same intelligence. Another example are humans who cannot hear and speak but are intelligent. They only have the problem to get the knowledge from other humans since language is the common social communication protocol to transfer knowledge from brain to brain. In my opinion language is overestimated in AI for the following reason: When we think we believe that we think in our language. From this we conclude that our thoughts are inherently structured by linguistic elements. And if our thoughts are so deeply connected with language then it is a small step to conclude that our whole intelligence depends inherently on language. But this is a misconception. We do not have conscious control over all of our thoughts. Most of the activities within our brain we cannot be aware of when we think. Nevertheless it is very useful and even essential for human intelligence being able to observe at least a subset of the own thoughts. It is this subset which we usually identify with the whole set of thoughts. But in fact it is just a tiny subset of all what happens in the 10^11 neurons. For the top-level observation of the own thoughts the brain uses the learned language. But this is no contradiction to the point that language is just a communication protocol and nothing else. The brain translates its patterns into language and routes this information to its own input regions. The reason why the brain uses language in order to observe its own thoughts is probably the following: If a person A wants to communicate some of its patterns to a person B then it has solve two problems: 1. How to compress the patterns? 2. How to send the patterns to the person B? The solution for the two problems is language. If a brain wants to observe its own thoughts it has to solve the same problems. The thoughts have to be compressed. If not you would observe every element of your thoughts and you would end up in an explosion of complexity. So why not use the same compression algorithm as it is used for communication with other people? That's the reason why the brain uses language when it observes its own thoughts. This phenomenon leads to the misconception that language is inherently connected with thoughts and intelligence. In fact it is just a top level communication protocol between two brains and within a single brain. Future AGI will have a much broader bandwidth and even for the current possibilities of technology human language would be a weak communication protocol for its internal observation of its own thoughts. - Matthias Terren Suydam wrote: Nice post. I'm not sure language is separable from any kind of intelligence we can meaningfully interact with. It's important to note (at least) two ways of talking about language: 1. specific aspects of language - what someone building an NLP module is focused on (e.g. the rules of English grammar and such). 2. the process of language - the expression of the internal state in some outward form in such a way that conveys shared meaning. If we conceptualize language as in #2, we can be talking about a great many human activities besides conversing: playing chess, playing music, programming computers, dancing, and so on. And in each example listed there is a learning curve that goes from pure novice to halting sufficiency to masterful fluency, just like learning a language. So *specific* forms of language (including the non-linguistic) are not in themselves important to intelligence (perhaps this is Matthias' point?), but the process of outwardly expressing meaning is fundamental to any social intelligence. Terren --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] META: A possible re-focusing of this list
This sounds good to me. I am much more drawn to topic #1. Topic #2 I have seen discussed recursively and in dozens of variants multiple places. The only thing I will add to Topic #2 is that I very seriously doubt current human intelligence individually or collectively is sufficient to address or meaningfully resolve or even crisply articulate such questions. Much more is accomplished by actually looking into the horse's mouth than philosophizing endlessly. - samantha Ben Goertzel wrote: Hi all, I have been thinking a bit about the nature of conversations on this list. It seems to me there are two types of conversations here: 1) Discussions of how to design or engineer AGI systems, using current computers, according to designs that can feasibly be implemented by moderately-sized groups of people 2) Discussions about whether the above is even possible -- or whether it is impossible because of weird physics, or poorly-defined special characteristics of human creativity, or the so-called complex systems problem, or because AGI intrinsically requires billions of people and quadrillions of dollars, or whatever Personally I am pretty bored with all the conversations of type 2. It's not that I consider them useless discussions in a grand sense ... certainly, they are valid topics for intellectual inquiry. But, to do anything real, you have to make **some** decisions about what approach to take, and I've decided long ago to take an approach of trying to engineer an AGI system. Now, if someone had a solid argument as to why engineering an AGI system is impossible, that would be important. But that never seems to be the case. Rather, what we hear are long discussions of peoples' intuitions and opinions in this regard. People are welcome to their own intuitions and opinions, but I get really bored scanning through all these intuitions about why AGI is impossible. One possibility would be to more narrowly focus this list, specifically on **how to make AGI work**. If this re-focusing were done, then philosophical arguments about the impossibility of engineering AGI in the near term would be judged **off topic** by definition of the list purpose. Potentially, there could be another list, something like agi-philosophy, devoted to philosophical and weird-physics and other discussions about whether AGI is possible or not. I am not sure whether I feel like running that other list ... and even if I ran it, I might not bother to read it very often. I'm interested in new, substantial ideas related to the in-principle possibility of AGI, but not interested at all in endless philosophical arguments over various peoples' intuitions in this regard. One fear I have is that people who are actually interested in building AGI, could be scared away from this list because of the large volume of anti-AGI philosophical discussion. Which, I add, almost never has any new content, and mainly just repeats well-known anti-AGI arguments (Penrose-like physics arguments ... mind is too complex to engineer, it has to be evolved ... no one has built an AGI yet therefore it will never be done ... etc.) What are your thoughts on this? -- Ben On Wed, Oct 15, 2008 at 10:49 AM, Jim Bromer [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote: On Wed, Oct 15, 2008 at 10:14 AM, Ben Goertzel [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote: Actually, I think COMP=false is a perfectly valid subject for discussion on this list. However, I don't think discussions of the form I have all the answers, but they're top-secret and I'm not telling you, hahaha are particularly useful. So, speaking as a list participant, it seems to me this thread has probably met its natural end, with this reference to proprietary weird-physics IP. However, speaking as list moderator, I don't find this thread so off-topic or unpleasant as to formally kill the thread. -- Ben If someone doesn't want to get into a conversation with Colin about whatever it is that he is saying, then they should just exercise some self-control and refrain from doing so. I think Colin's ideas are pretty far out there. But that does not mean that he has never said anything that might be useful. My offbeat topic, that I believe that the Lord may have given me some direction about a novel approach to logical satisfiability that I am working on, but I don't want to discuss the details about the algorithms until I have gotten a chance to see if they work or not, was never intended to be a discussion about the theory itself. I wanted to have a discussion about whether or not a good SAT solution would have a significant influence on AGI, and whether or not the unlikely discovery of an unexpected breakthrough on SAT would serve as rational evidence in support of the theory
Re: RSI without input (was Re: [agi] Updated AGI proposal (CMR v2.1))
Matt Mahoney wrote: --- On Tue, 10/14/08, Charles Hixson [EMAIL PROTECTED] wrote: It seems clear that without external inputs the amount of improvement possible is stringently limited. That is evident from inspection. But why the without input? The only evident reason is to ensure the truth of the proposition, as it doesn't match any intended real-world scenario that I can imagine. (I've never considered the Oracle AI scenario [an AI kept within a black box that will answer all your questions without inputs] to be plausible.) If input is allowed, then we can't clearly distinguish between self improvement and learning. Clearly, learning is a legitimate form of improvement, but it is not *self* improvement. What I am trying to debunk is the perceived risk of a fast takeoff singularity launched by the first AI to achieve superhuman intelligence. In this scenario, a scientist with an IQ of 180 produces an artificial scientist with an IQ of 200, which produces an artificial scientist with an IQ of 250, and so on. I argue it can't happen because human level intelligence is the wrong threshold. There is currently a global brain (the world economy) with an IQ of around 10^10, and approaching 10^12. Oh man. It is so tempting in today's economic morass to point out the obvious stupidity of this purported super-super-genius. Why would you assign such an astronomical intelligence to the economy? Even from the POV of the best of Austrian micro-economic optimism it is not at all clear that billions of minds of human level IQ interacting with one another can be said to produce some such large exponential of the average human IQ.How much of the advancement of humanity is the result of a relatively few exceptionally bright minds rather than the billions of lesser intelligences? Are you thinking more of the entire cultural environment rather than specifically the economy? - samantha --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] constructivist issues
Abram, I find it more useful to think in terms of Chaitin's reformulation of Godel's Theorem: http://www.cs.auckland.ac.nz/~chaitin/sciamer.html Given any computer program with algorithmic information capacity less than K, it cannot prove theorems whose algorithmic information content is greater than K. Put simply, there are some things our brains are not big enough to prove true or false This is true for quantum computers just as it's true for classical computers. Penrose hypothesized it would NOT hold for quantum gravity computers, but IMO this is a fairly impotent hypothesis because quantum gravity computers don't exist (even theoretically, I mean: since there is no unified quantum gravity theory yet). Penrose assumes that humans don't have this sort of limitation, but I'm not sure why. On the other hand, this limitation can be overcome somewhat if you allow the program P to interact with the external world in a way that lets it be modified into P1 such that P1 is not computable by P. In this case P needs to have a guru (or should I say an oracle ;-) that it trusts to modify itself in ways it can't understand, or else to be a gambler-type... You seem almost confused when you say that an AI can't reason about uncomputable entities. Of course it can. An AI can manipulate math symbols in a certain formal system, and then associate these symbols with the words uncomputable entities, and with its own self ... or us. This is what we do. An AI program can't actually manipulate the uncomputable entities directly , but what makes you think *we* can, either? -- Ben G On Sat, Oct 18, 2008 at 9:54 PM, Abram Demski [EMAIL PROTECTED] wrote: Matt, I suppose you don't care about Steve's do not comment request? Oh well, I want to discuss this anyway. 'Tis why I posted in the first place. No, I do not claim that computer theorem-provers cannot prove Goedel's Theorem. It has been done. The objection applies specifically to AIXI-- AIXI cannot prove goedel's theorem. More generally, all AIXI's world-models are computable. What do I mean when I say to reason about non-computable entities? Well, Goedel's Incompleteness Theorem is a fine example. Another example is the way humans can talk about whether a particular program will halt. This sort of thing can be done in logical systems by adding basic non-computable primitives. A common choice is to add numbers. (Numbers may seem like the prototypical computable thing, but any logic of numbers is incomplete, as Goedel showed of course.) The broader issue is that *in general* given any ideal model of intelligence similar to AIXI, with a logically defined class of world-models, it will be possible to point out something that the intelligence cannot possibly reason about-- namely, its own semantics. This follows from Tarski's indefinability theorem, and hinges on a few assumptions about the meaning of logically defined. I am not altogether sure that the Novamente/OCP design is really an approximation of AIXI anyway, *but* I think it is a serious concern. If the Novamente/OCP design really solves the (broader) problem, then it also solves some key problems in epistemology (specifically, in formal theories of truth), so it would be very interesting to see it worked out in these terms. --Abram On Sat, Oct 18, 2008 at 9:12 PM, Matt Mahoney [EMAIL PROTECTED] wrote: --- On Sat, 10/18/08, Abram Demski [EMAIL PROTECTED] wrote: Non-Constructive Logic: Any AI method that approximates AIXI will lack the human capability to reason about non-computable entities. Then how is it that humans can do it? According to the AIXI theorem, if we can do this, it makes us less able to achieve our goals because AIXI is provably optimal. Exactly what do you mean by reason about non-computable entities? Do you claim that a computer could not discover a proof of Goedel's incompleteness theorem by brute force search? -- Matt Mahoney, [EMAIL PROTECTED] --- 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/?; Powered by Listbox: http://www.listbox.com -- Ben Goertzel, PhD CEO, Novamente LLC and Biomind LLC Director of Research, SIAI [EMAIL PROTECTED] Nothing will ever be attempted if all possible objections must be first overcome - Dr Samuel Johnson --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription:
AW: AW: [agi] Re: Defining AGI
What the computer makes with the data it receives depends on the information of the transferred data, its internal algorithms and its internal data. This is the same with humans and natural language. Language understanding would be useful to teach the AGI with existing knowledge already represented in natural language. But natural language understanding suffers from the problem of ambiguities. These ambiguities can be solved by having similar knowledge as humans have. But then you have a recursive problem because first there has to be solved the problem to obtain this knowledge. Nature solves this problem with embodiment. Different people make similar experiences since the laws of nature do not depend on space and time. Therefore we all can imagine a dog which is angry. Since we have experienced angry dogs but we haven't experienced angry trees we can resolve the linguistic ambiguity of my former example and answer the question: Who was angry? The way to obtain knowledge with embodiment is hard and long even in virtual worlds. If the AGI shall understand natural language it would be necessary that it makes similar experiences as humans make in the real world. But this would need a very very sophisticated and rich virtual world. At least, there have to be angry dogs in the virtual world ;-) As I have already said I do not think the relation between utility of this approach and the costs would be positive for first AGI. - Matthias William Pearson [mailto:[EMAIL PROTECTED] wrote If I specify in a language to a computer that it should do something, it will do it no matter what (as long as I have sufficient authority). Telling a human to do something, e.g. wave your hands in the air and shout, the human will decide to do that based on how much it trusts you and whether they think it is a good idea. Generally a good idea in a situation where you are attracting the attention of rescuers, otherwise likely to make you look silly. I'm generally in favour of getting some NLU into AIs mainly because a lot of the information we have about the world is still in that form, so an AI without access to that information would have to reinvent it, which I think would take a long time. Even mathematical proofs are still somewhat in natural language. Other than that you could work on machine language understanding where information was taken in selectively and judged on its merits not its security credentials. Will Pearson --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: [agi] Re: Defining AGI
An excellent post, thanks! IMO, it raises the bar for discussion of language and AGI, and should be carefully considered by the authors of future posts on the topic of language and AGI. If the AGI list were a forum, Matthias's post should be pinned! -dave On Sun, Oct 19, 2008 at 6:58 PM, Dr. Matthias Heger [EMAIL PROTECTED] wrote: The process of outwardly expressing meaning may be fundamental to any social intelligence but the process itself needs not much intelligence. Every email program can receive meaning, store meaning and it can express it outwardly in order to send it to another computer. It even can do it without loss of any information. Regarding this point, it even outperforms humans already who have no conscious access to the full meaning (information) in their brains. The only thing which needs much intelligence from the nowadays point of view is the learning of the process of outwardly expressing meaning, i.e. the learning of language. The understanding of language itself is simple. To show that intelligence is separated from language understanding I have already given the example that a person could have spoken with Einstein but needed not to have the same intelligence. Another example are humans who cannot hear and speak but are intelligent. They only have the problem to get the knowledge from other humans since language is the common social communication protocol to transfer knowledge from brain to brain. In my opinion language is overestimated in AI for the following reason: When we think we believe that we think in our language. From this we conclude that our thoughts are inherently structured by linguistic elements. And if our thoughts are so deeply connbected with language then it is a small step to conclude that our whole intelligence depends inherently on language. But this is a misconception. We do not have conscious control over all of our thoughts. Most of the activities within our brain we cannot be aware of when we think. Nevertheless it is very useful and even essential for human intelligence being able to observe at least a subset of the own thoughts. It is this subset which we usually identify with the whole set of thoughts. But in fact it is just a tiny subset of all what happens in the 10^11 neurons. For the top-level observation of the own thoughts the brain uses the learned language. But this is no contradiction to the point that language is just a communication protocol and nothing else. The brain translates its patterns into language and routes this information to its own input regions. The reason why the brain uses language in order to observe its own thoughts is probably the following: If a person A wants to communicate some of its patterns to a person B then it has solve two problems: 1. How to compress the patterns? 2. How to send the patterns to the person B? The solution for the two problems is language. If a brain wants to observe its own thoughts it has to solve the same problems. The thoughts have to be compressed. If not you would observe every element of your thoughts and you would end up in an explosion of complexity. So why not use the same compression algorithm as it is used for communication with other people? That's the reason why the brain uses language when it observes its own thoughts. This phenomenon leads to the misconception that language is inherently connected with thoughts and intelligence. In fact it is just a top level communication protocol between two brains and within a single brain. Future AGI will have a much broader bandwidth and even for the current possibilities of technology human language would be a weak communication protocol for its internal observation of its own thoughts. - Matthias --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: [agi] Re: Defining AGI
2008/10/19 Dr. Matthias Heger [EMAIL PROTECTED]: The process of outwardly expressing meaning may be fundamental to any social intelligence but the process itself needs not much intelligence. Every email program can receive meaning, store meaning and it can express it outwardly in order to send it to another computer. It even can do it without loss of any information. Regarding this point, it even outperforms humans already who have no conscious access to the full meaning (information) in their brains. The only thing which needs much intelligence from the nowadays point of view is the learning of the process of outwardly expressing meaning, i.e. the learning of language. The understanding of language itself is simple. I'd disagree, there is another part of dealing with language that we don't have a good idea of how to do. Deciding whether to assimilate it and if so how. If I specify in a language to a computer that it should do something, it will do it no matter what (as long as I have sufficient authority). Telling a human to do something, e.g. wave your hands in the air and shout, the human will decide to do that based on how much it trusts you and whether they think it is a good idea. Generally a good idea in a situation where you are attracting the attention of rescuers, otherwise likely to make you look silly. I'm generally in favour of getting some NLU into AIs mainly because a lot of the information we have about the world is still in that form, so an AI without access to that information would have to reinvent it, which I think would take a long time. Even mathematical proofs are still somewhat in natural language. Other than that you could work on machine language understanding where information was taken in selectively and judged on its merits not its security credentials. Will Pearson --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] Re: Meaning, communication and understanding
regarding denotational semantics: I prefer to think of the meaning of X as the fuzzy set of patterns associated with X. (In fact, I recall giving a talk on this topic at a meeting of the American Math Society in 1990 ;-) On Sun, Oct 19, 2008 at 6:59 AM, Vladimir Nesov [EMAIL PROTECTED] wrote: On Sun, Oct 19, 2008 at 11:58 AM, Dr. Matthias Heger [EMAIL PROTECTED] wrote: The process of outwardly expressing meaning may be fundamental to any social intelligence but the process itself needs not much intelligence. Every email program can receive meaning, store meaning and it can express it outwardly in order to send it to another computer. It even can do it without loss of any information. Regarding this point, it even outperforms humans already who have no conscious access to the full meaning (information) in their brains. The only thing which needs much intelligence from the nowadays point of view is the learning of the process of outwardly expressing meaning, i.e. the learning of language. The understanding of language itself is simple. Meaning is tricky business. As far as I can tell, meaning Y of a system X is an external model that relates system X to its meaning Y (where meaning may be a physical object, or a class of objects, where each individual object figures into the model). Formal semantics works this way (see http://en.wikipedia.org/wiki/Denotational_semantics ). When you are thinking about an object, the train of though depends on your experience about that object, and will influence your behavior in situations depending on information about that objects. Meaning propagates through the system according to rules of the model, propagates inferentially in the model and not in the system, and so can reach places and states of the system not at all obviously concerned with what this semantic model relates them to. And conversely, meaning doesn't magically appear where model doesn't say it does: if system is broken, meaning is lost, at least until you come up with another model and relate it to the previous one. When you say that e-mail contains meaning and network transfers meaning, it is an assertion about the model of content of e-mail, that relates meaning in the mind of the writer to bits in the memory of machines. From this point of view, we can legitemately say that meaning is transferred, and is expressed. But the same meaning doesn't exist in e-mails if you cut them from the mind that expressed the meaning in the form of e-mails, or experience that transferred meaning in the mind. Understanding is the process of integrating different models, different meanings, different pieces of information as seen by your model. It is the ability to translate pieces of information that have nontrivial structure, in your basis. Normal use of understanding applies only to humans, everything else generalizes this concept in sometimes very strange ways. When we say that person understood something, in this language it's equivalent to person having successfully integrated that piece in his mind, our model of that person starting to attribute properties of that piece of information to his thought and behavior. So, you are cutting this knot at a trivial point. The difficulty is in the translation, but you point on one side of the translation process and say that this side is simple, then point to another than say that this side is hard. The problem is that it's hard to put a finger on the point just after translation, but it's easy to see how our technology, as physical medium, transfers information ready for translation. This outward appearance has little bearing on semantic models. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.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/?; Powered by Listbox: http://www.listbox.com -- Ben Goertzel, PhD CEO, Novamente LLC and Biomind LLC Director of Research, SIAI [EMAIL PROTECTED] Nothing will ever be attempted if all possible objections must be first overcome - Dr Samuel Johnson --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
[agi] Re: Meaning, communication and understanding
On Sun, Oct 19, 2008 at 11:58 AM, Dr. Matthias Heger [EMAIL PROTECTED] wrote: The process of outwardly expressing meaning may be fundamental to any social intelligence but the process itself needs not much intelligence. Every email program can receive meaning, store meaning and it can express it outwardly in order to send it to another computer. It even can do it without loss of any information. Regarding this point, it even outperforms humans already who have no conscious access to the full meaning (information) in their brains. The only thing which needs much intelligence from the nowadays point of view is the learning of the process of outwardly expressing meaning, i.e. the learning of language. The understanding of language itself is simple. Meaning is tricky business. As far as I can tell, meaning Y of a system X is an external model that relates system X to its meaning Y (where meaning may be a physical object, or a class of objects, where each individual object figures into the model). Formal semantics works this way (see http://en.wikipedia.org/wiki/Denotational_semantics ). When you are thinking about an object, the train of though depends on your experience about that object, and will influence your behavior in situations depending on information about that objects. Meaning propagates through the system according to rules of the model, propagates inferentially in the model and not in the system, and so can reach places and states of the system not at all obviously concerned with what this semantic model relates them to. And conversely, meaning doesn't magically appear where model doesn't say it does: if system is broken, meaning is lost, at least until you come up with another model and relate it to the previous one. When you say that e-mail contains meaning and network transfers meaning, it is an assertion about the model of content of e-mail, that relates meaning in the mind of the writer to bits in the memory of machines. From this point of view, we can legitemately say that meaning is transferred, and is expressed. But the same meaning doesn't exist in e-mails if you cut them from the mind that expressed the meaning in the form of e-mails, or experience that transferred meaning in the mind. Understanding is the process of integrating different models, different meanings, different pieces of information as seen by your model. It is the ability to translate pieces of information that have nontrivial structure, in your basis. Normal use of understanding applies only to humans, everything else generalizes this concept in sometimes very strange ways. When we say that person understood something, in this language it's equivalent to person having successfully integrated that piece in his mind, our model of that person starting to attribute properties of that piece of information to his thought and behavior. So, you are cutting this knot at a trivial point. The difficulty is in the translation, but you point on one side of the translation process and say that this side is simple, then point to another than say that this side is hard. The problem is that it's hard to put a finger on the point just after translation, but it's easy to see how our technology, as physical medium, transfers information ready for translation. This outward appearance has little bearing on semantic models. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] Re: Meaning, communication and understanding
On Sun, Oct 19, 2008 at 3:09 PM, Ben Goertzel [EMAIL PROTECTED] wrote: regarding denotational semantics: I prefer to think of the meaning of X as the fuzzy set of patterns associated with X. (In fact, I recall giving a talk on this topic at a meeting of the American Math Society in 1990 ;-) I like denotational semantics as an example (even though it doesn't suggest uncertainty), because it's a well-understood semantic model with meaning assigned to deep intermediate steps, in nontrivial ways. It's easier to see by analogy to this how abstract thought that relates to misremembered experience of 20 years ago and that never gets outwardly expressed still has meaning, and how to assign it which meaning. What form meaning takes depends on the model that assigns meaning to the system, which when we cross the line into realm of human-level understanding becomes a mind, and so meaning, in a technical sense, becomes a functional aspect of AGI. If AGI works on something called fuzzy set of patterns, then it's the meaning of what it models. There is of course a second step when you yourself, as an engineer, assign meaning to aspects of operation of AGI, and to relations between AGI and what it models, in your own head, but this perspective loses technical precision, although to some extent it's necessary. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
[agi] Words vs Concepts [ex Defining AGI]
Matthias, You seem - correct me - to be going a long way round saying that words are different from concepts - they're just sound-and-letter labels for concepts, which have a very different form. And the processing of words/language is distinct from and relatively simple compared to the processing of the underlying concepts. So take THE CAT SAT ON THE MAT or THE MIND HAS ONLY CERTAIN PARTS WHICH ARE SENTIENT or THE US IS THE HOME OF THE FINANCIAL CRISIS the words c-a-t or m-i-n-d or U-S or f-i-n-a-n-c-i-a-l c-r-i-s-i-s are distinct from the underlying concepts. The question is: What form do those concepts take? And what is happening in our minds (and what has to happen in any mind) when we process those concepts? You talk of patterns. What patterns, do you think, form the concept of mind that are engaged in thinking about sentence 2? Do you think that concepts like mind or the US might involve something much more complex still? Models? Or is that still way too simple? Spaces? Equally, of course, we can say that each *sentence* above is not just a verbal composition but a conceptual composition - and the question then is what form does such a composition take? Do sentences form, say, a pattern of patterns, or something like a picture? Or a blending of spaces ? Or are concepts like *money*? YOU CAN BUY A LOT WITH A MILLION DOLLARS Does every concept function somewhat like money, e.g. a million dollars - something that we know can be cashed in, in an infinite variety of ways, but that we may not have to start cashing in, (when processing), unless really called for - or only cash in so far? P.S. BTW this is the sort of psycho-philosophical discussion that I would see as central to AGI, but that most of you don't want to talk about? Matthias: What the computer makes with the data it receives depends on the information of the transferred data, its internal algorithms and its internal data. This is the same with humans and natural language. Language understanding would be useful to teach the AGI with existing knowledge already represented in natural language. But natural language understanding suffers from the problem of ambiguities. These ambiguities can be solved by having similar knowledge as humans have. But then you have a recursive problem because first there has to be solved the problem to obtain this knowledge. Nature solves this problem with embodiment. Different people make similar experiences since the laws of nature do not depend on space and time. Therefore we all can imagine a dog which is angry. Since we have experienced angry dogs but we haven't experienced angry trees we can resolve the linguistic ambiguity of my former example and answer the question: Who was angry? The way to obtain knowledge with embodiment is hard and long even in virtual worlds. If the AGI shall understand natural language it would be necessary that it makes similar experiences as humans make in the real world. But this would need a very very sophisticated and rich virtual world. At least, there have to be angry dogs in the virtual world ;-) As I have already said I do not think the relation between utility of this approach and the costs would be positive for first AGI. --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
AW: [agi] Re: Meaning, communication and understanding
I agree that understanding is the process of integrating different models, different meanings, different pieces of information as seen by your model. But this integrating just matching and not extending the own model with new entities. You only match linguistic entities of received linguistically represented information with existing entities of your model (i.e. with some of your existing patterns). If you could manage the matching process successfully then you have understood the linguistic message. Natural communication and language understanding is completely comparable with common processes in computer science. There is an internal data representation. A subset of this data is translated into a linguistic string and transferred to another agent which retranslates the message before it possibly but not necessarily changes its database. The only reason why natural language understanding is so difficult is because it needs a lot of knowledge to resolve ambiguities which humans usually gain via own experience. But alone from being able to resolve the ambiguities and being able to do the matching process successfully you will know nothing about the creation of patterns and the way how to work intelligently with these patterns. Therefore communication is separated from these main problems of AGI in the same way as communication is completely separated from the structure and algorithms of the database of computers. Only the process of *learning* such a communication would be AI (I am not sure if it is AGI). But you cannot learn to communicate if there is nothing to communicate. So every approach towards AGI via *learning* language understanding will need at least a further domain for the content of communication. Probably you need even more domains because the linguistic ambiguities can resolved only with broad knowledge . And this is my point why I say that language understanding would yield costs which are not necessary. We can build AGI just by concentrating all efforts to a *single* domain with very useful properties (i.e. domain of mathematics). This would reduce the immense costs of simulating real worlds and additionally concentrating on *at least two* domains at the same time. -Matthias Vladimir Nesov [mailto:[EMAIL PROTECTED] wrote Gesendet: Sonntag, 19. Oktober 2008 12:59 An: agi@v2.listbox.com Betreff: [agi] Re: Meaning, communication and understanding On Sun, Oct 19, 2008 at 11:58 AM, Dr. Matthias Heger [EMAIL PROTECTED] wrote: The process of outwardly expressing meaning may be fundamental to any social intelligence but the process itself needs not much intelligence. Every email program can receive meaning, store meaning and it can express it outwardly in order to send it to another computer. It even can do it without loss of any information. Regarding this point, it even outperforms humans already who have no conscious access to the full meaning (information) in their brains. The only thing which needs much intelligence from the nowadays point of view is the learning of the process of outwardly expressing meaning, i.e. the learning of language. The understanding of language itself is simple. Meaning is tricky business. As far as I can tell, meaning Y of a system X is an external model that relates system X to its meaning Y (where meaning may be a physical object, or a class of objects, where each individual object figures into the model). Formal semantics works this way (see http://en.wikipedia.org/wiki/Denotational_semantics ). When you are thinking about an object, the train of though depends on your experience about that object, and will influence your behavior in situations depending on information about that objects. Meaning propagates through the system according to rules of the model, propagates inferentially in the model and not in the system, and so can reach places and states of the system not at all obviously concerned with what this semantic model relates them to. And conversely, meaning doesn't magically appear where model doesn't say it does: if system is broken, meaning is lost, at least until you come up with another model and relate it to the previous one. When you say that e-mail contains meaning and network transfers meaning, it is an assertion about the model of content of e-mail, that relates meaning in the mind of the writer to bits in the memory of machines. From this point of view, we can legitemately say that meaning is transferred, and is expressed. But the same meaning doesn't exist in e-mails if you cut them from the mind that expressed the meaning in the form of e-mails, or experience that transferred meaning in the mind. Understanding is the process of integrating different models, different meanings, different pieces of information as seen by your model. It is the ability to translate pieces of information that have nontrivial structure, in your basis. Normal use of understanding applies only to humans, everything else generalizes this
Re: [agi] Re: Meaning, communication and understanding
On Sun, Oct 19, 2008 at 5:23 PM, Dr. Matthias Heger [EMAIL PROTECTED] wrote: I agree that understanding is the process of integrating different models, different meanings, different pieces of information as seen by your model. But this integrating just matching and not extending the own model with new entities. You only match linguistic entities of received linguistically represented information with existing entities of your model (i.e. with some of your existing patterns). If you could manage the matching process successfully then you have understood the linguistic message. Natural communication and language understanding is completely comparable with common processes in computer science. There is an internal data representation. A subset of this data is translated into a linguistic string and transferred to another agent which retranslates the message before it possibly but not necessarily changes its database. The only reason why natural language understanding is so difficult is because it needs a lot of knowledge to resolve ambiguities which humans usually gain via own experience. But alone from being able to resolve the ambiguities and being able to do the matching process successfully you will know nothing about the creation of patterns and the way how to work intelligently with these patterns. Therefore communication is separated from these main problems of AGI in the same way as communication is completely separated from the structure and algorithms of the database of computers. Only the process of *learning* such a communication would be AI (I am not sure if it is AGI). But you cannot learn to communicate if there is nothing to communicate. So every approach towards AGI via *learning* language understanding will need at least a further domain for the content of communication. Probably you need even more domains because the linguistic ambiguities can resolved only with broad knowledge . And this is my point why I say that language understanding would yield costs which are not necessary. We can build AGI just by concentrating all efforts to a *single* domain with very useful properties (i.e. domain of mathematics). This would reduce the immense costs of simulating real worlds and additionally concentrating on *at least two* domains at the same time. I think I see what you are trying to communicate. Correct me if I got something wrong here. You assume a certain architectural decision for AIs in question when you talk about this interpretation of process of communication. Basically, AI1 communicates with AI2, and they both work with two domains: D and L, D being internal domain and L being communication domain, stuff that gets sent via e-mail. AI1 translates meaning D1 into message L1, which is transferred as L2 to AI2, which then translates it to D2. You call a step L2-D2 understanding or matching, also assuming that this process doesn't need to change AI2, to make it change its model, to learn. You then suggest that L doesn't need to be natural language, as D for language is the most difficult real world, and then instead we need to pick easier L and D and work on their interplay. If AI1 already can translate between D and L, AI2 might need to learn to translate between L and D on its own, knowing only D at the start, and this ability you suggest as central challenge of intelligence. I think that this model is overly simplistic, overemphasizing an artificial divide between domains within AI's cognition (L and D), and externalizing communication domain from the core of AI. Both world model and language model support interaction with environment, there is no clear cognitive distinction between them. As a given, interaction happens at the narrow I/O interface, and anything else is a design decision for a specific AI (even invariability of I/O is, a simplifying assumption that complicates semantics of time and more radical self-improvement). Sufficiently flexible cognitive algorithm should be able to integrate facts about any domain, becoming able to generate appropriate behavior in corresponding contexts. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
AW: [agi] Words vs Concepts [ex Defining AGI]
For the discussion of the subject the details of the pattern representation are not important at all. It is sufficient if you agree that a spoken sentence represent a certain set of patterns which are translated into the sentence. The receiving agent retranslates the sentence and matches the content with its model by activating similar patterns. The activation of patterns is extremely fast and happens in real time. The brain even predicts patterns if it just hears the first syllable of a word: http://www.rochester.edu/news/show.php?id=3244 There is no creation of new patterns and there is no intelligent algorithm which manipulates patterns. It is just translating, sending, receiving and retranslating. From the ambiguities of natural language you obtain some hints about the structure of the patterns. But you cannot even expect to obtain all detail of these patterns by understanding the process of language understanding. There will be probably many details within these patterns which are only necessary for internal calculations. These details will be not visible from the linguistic point of view. Just think about communicating computers and you will know what I mean. - Matthias Mike Tintner [mailto:[EMAIL PROTECTED] wrote: Matthias, You seem - correct me - to be going a long way round saying that words are different from concepts - they're just sound-and-letter labels for concepts, which have a very different form. And the processing of words/language is distinct from and relatively simple compared to the processing of the underlying concepts. So take THE CAT SAT ON THE MAT or THE MIND HAS ONLY CERTAIN PARTS WHICH ARE SENTIENT or THE US IS THE HOME OF THE FINANCIAL CRISIS the words c-a-t or m-i-n-d or U-S or f-i-n-a-n-c-i-a-l c-r-i-s-i-s are distinct from the underlying concepts. The question is: What form do those concepts take? And what is happening in our minds (and what has to happen in any mind) when we process those concepts? You talk of patterns. What patterns, do you think, form the concept of mind that are engaged in thinking about sentence 2? Do you think that concepts like mind or the US might involve something much more complex still? Models? Or is that still way too simple? Spaces? Equally, of course, we can say that each *sentence* above is not just a verbal composition but a conceptual composition - and the question then is what form does such a composition take? Do sentences form, say, a pattern of patterns, or something like a picture? Or a blending of spaces ? Or are concepts like *money*? YOU CAN BUY A LOT WITH A MILLION DOLLARS Does every concept function somewhat like money, e.g. a million dollars - something that we know can be cashed in, in an infinite variety of ways, but that we may not have to start cashing in, (when processing), unless really called for - or only cash in so far? P.S. BTW this is the sort of psycho-philosophical discussion that I would see as central to AGI, but that most of you don't want to talk about? Matthias: What the computer makes with the data it receives depends on the information of the transferred data, its internal algorithms and its internal data. This is the same with humans and natural language. Language understanding would be useful to teach the AGI with existing knowledge already represented in natural language. But natural language understanding suffers from the problem of ambiguities. These ambiguities can be solved by having similar knowledge as humans have. But then you have a recursive problem because first there has to be solved the problem to obtain this knowledge. Nature solves this problem with embodiment. Different people make similar experiences since the laws of nature do not depend on space and time. Therefore we all can imagine a dog which is angry. Since we have experienced angry dogs but we haven't experienced angry trees we can resolve the linguistic ambiguity of my former example and answer the question: Who was angry? The way to obtain knowledge with embodiment is hard and long even in virtual worlds. If the AGI shall understand natural language it would be necessary that it makes similar experiences as humans make in the real world. But this would need a very very sophisticated and rich virtual world. At least, there have to be angry dogs in the virtual world ;-) As I have already said I do not think the relation between utility of this approach and the costs would be positive for first AGI. --- 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
Re: [agi] constructivist issues
Ben, I don't know what sounded almost confused, but anyway it is apparent that I didn't make my position clear. I am not saying we can manipulate these things directly via exotic (non)computing. First, I am very specifically saying that AIXI-style AI (meaning, any AI that approaches AIXI as resources increase) cannot reason about uncomputable entities. This is because AIXI entertains only computable models. Second, I am suggesting a broader problem that will apply to a wide class of formulations of idealized intelligence such as AIXI: if their internal logic obeys a particular set of assumptions, it will become prone to Tarski's Undefinability Theorem. Therefore, we humans will be able to point out a particular class of concepts that it cannot reason about; specifically, the very concepts used in describing the ideal intelligence in the first place. One reasonable way of avoiding the humans are magic explanation of this (or humans use quantum gravity computing, etc) is to say that, OK, humans really are an approximation of an ideal intelligence obeying those assumptions. Therefore, we cannot understand the math needed to define our own intelligence. Therefore, we can't engineer human-level AGI. I don't like this conclusion! I want a different way out. I'm not sure the guru explanation is enough... who was the Guru for Humankind? Thanks, --Abram On Sun, Oct 19, 2008 at 5:39 AM, Ben Goertzel [EMAIL PROTECTED] wrote: Abram, I find it more useful to think in terms of Chaitin's reformulation of Godel's Theorem: http://www.cs.auckland.ac.nz/~chaitin/sciamer.html Given any computer program with algorithmic information capacity less than K, it cannot prove theorems whose algorithmic information content is greater than K. Put simply, there are some things our brains are not big enough to prove true or false This is true for quantum computers just as it's true for classical computers. Penrose hypothesized it would NOT hold for quantum gravity computers, but IMO this is a fairly impotent hypothesis because quantum gravity computers don't exist (even theoretically, I mean: since there is no unified quantum gravity theory yet). Penrose assumes that humans don't have this sort of limitation, but I'm not sure why. On the other hand, this limitation can be overcome somewhat if you allow the program P to interact with the external world in a way that lets it be modified into P1 such that P1 is not computable by P. In this case P needs to have a guru (or should I say an oracle ;-) that it trusts to modify itself in ways it can't understand, or else to be a gambler-type... You seem almost confused when you say that an AI can't reason about uncomputable entities. Of course it can. An AI can manipulate math symbols in a certain formal system, and then associate these symbols with the words uncomputable entities, and with its own self ... or us. This is what we do. An AI program can't actually manipulate the uncomputable entities directly , but what makes you think *we* can, either? -- Ben G --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] Words vs Concepts [ex Defining AGI]
Matthias, I take the point that there is vastly more to language understanding than the surface processing of words as opposed to concepts. I agree that it is typically v. fast. I don't think though that you can call any concept a pattern. On the contrary, a defining property of concepts, IMO, is that they resist reduction to any pattern or structure - which is rather important, since my impression is most AGI-ers live by patterns/structures. Even a concept like triangle cannot actually be reduced to a pattern. Try it, if you wish. And the issue of conceptualisation - of what a concept consists of - is manifestly an unsolved problem for both cog sci and AI and of utmost, central importance for AGI. We have to understand how the brain performs its feats here, because that, at a rough general level, is almost certainly how it will *have* to be done. (I can't resist being snide here and saying that since this an unsolved problem, one can virtually guarantee that AGI-ers will therefore refuse to discuss it). Trying to work out what information the brain handles, for example, when it talks about THE US IS THE HOME OF THE FINANCIAL CRISIS - what passes - and has to pass - through a mind thinking specifically of the financial crisis?- is in some ways as great a challenge as working out what the brain's engrams consist of. Clearly it won't be the kind of mere, symbolic, dictionary processing that some AGI-ers envisage. It will be perhaps as complex as the conceptualisation of party in: HOW WAS THE PARTY LAST NIGHT? where a single word may be used to touch upon over, say, two hours of sensory, movie experience in the brain. I partly disagree with you about how we should study all this - it is vital to look at how we understand, or rather fail to understand and get confused by concepts and language - which happens all the time. This can tell us a great deal about what is going on underneath. Matthias: For the discussion of the subject the details of the pattern representation are not important at all. It is sufficient if you agree that a spoken sentence represent a certain set of patterns which are translated into the sentence. The receiving agent retranslates the sentence and matches the content with its model by activating similar patterns. The activation of patterns is extremely fast and happens in real time. The brain even predicts patterns if it just hears the first syllable of a word: http://www.rochester.edu/news/show.php?id=3244 There is no creation of new patterns and there is no intelligent algorithm which manipulates patterns. It is just translating, sending, receiving and retranslating. From the ambiguities of natural language you obtain some hints about the structure of the patterns. But you cannot even expect to obtain all detail of these patterns by understanding the process of language understanding. There will be probably many details within these patterns which are only necessary for internal calculations. These details will be not visible from the linguistic point of view. Just think about communicating computers and you will know what I mean. - Matthias Mike Tintner [mailto:[EMAIL PROTECTED] wrote: Matthias, You seem - correct me - to be going a long way round saying that words are different from concepts - they're just sound-and-letter labels for concepts, which have a very different form. And the processing of words/language is distinct from and relatively simple compared to the processing of the underlying concepts. So take THE CAT SAT ON THE MAT or THE MIND HAS ONLY CERTAIN PARTS WHICH ARE SENTIENT or THE US IS THE HOME OF THE FINANCIAL CRISIS the words c-a-t or m-i-n-d or U-S or f-i-n-a-n-c-i-a-l c-r-i-s-i-s are distinct from the underlying concepts. The question is: What form do those concepts take? And what is happening in our minds (and what has to happen in any mind) when we process those concepts? You talk of patterns. What patterns, do you think, form the concept of mind that are engaged in thinking about sentence 2? Do you think that concepts like mind or the US might involve something much more complex still? Models? Or is that still way too simple? Spaces? Equally, of course, we can say that each *sentence* above is not just a verbal composition but a conceptual composition - and the question then is what form does such a composition take? Do sentences form, say, a pattern of patterns, or something like a picture? Or a blending of spaces ? Or are concepts like *money*? YOU CAN BUY A LOT WITH A MILLION DOLLARS Does every concept function somewhat like money, e.g. a million dollars - something that we know can be cashed in, in an infinite variety of ways, but that we may not have to start cashing in, (when processing), unless really called for - or only cash in so far? P.S. BTW this is the sort of psycho-philosophical discussion that I would see as central to AGI, but that most of you don't want to
Re: AW: AW: [agi] Re: Defining AGI
--- On Sun, 10/19/08, Dr. Matthias Heger [EMAIL PROTECTED] wrote: Every email program can receive meaning, store meaning and it can express it outwardly in order to send it to another computer. It even can do it without loss of any information. Regarding this point, it even outperforms humans already who have no conscious access to the full meaning (information) in their brains. Email programs do not store meaning, they store data. The email program has no understanding of the stuff it stores, so this is a poor analogy. The only thing which needs much intelligence from the nowadays point of view is the learning of the process of outwardly expressing meaning, i.e. the learning of language. The understanding of language itself is simple. Isn't the *learning* of language the entire point? If you don't have an answer for how an AI learns language, you haven't solved anything. The understanding of language only seems simple from the point of view of a fluent speaker. Fluency however should not be confused with a lack of intellectual effort - rather, it's a state in which the effort involved is automatic and beyond awareness. To show that intelligence is separated from language understanding I have already given the example that a person could have spoken with Einstein but needed not to have the same intelligence. Another example are humans who cannot hear and speak but are intelligent. They only have the problem to get the knowledge from other humans since language is the common social communication protocol to transfer knowledge from brain to brain. Einstein had to express his (non-linguistic) internal insights in natural language and in mathematical language. In both modalities he had to use his intelligence to make the translation from his mental models. Deaf people speak in sign language, which is only different from spoken language in superficial ways. This does not tell us much about language that we didn't already know. In my opinion language is overestimated in AI for the following reason: When we think we believe that we think in our language. From this we conclude that our thoughts are inherently structured by linguistic elements. And if our thoughts are so deeply connected with language then it is a small step to conclude that our whole intelligence depends inherently on language. It is surely true that much/most of our cognitive processing is not at all linguistic, and that there is much that happens beyond our awareness. However, language is a necessary tool, for humans at least, to obtain a competent conceptual framework, even if that framework ultimately transcends the linguistic dynamics that helped develop it. Without language it is hard to see how humans could develop self-reflectivity. Terren __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: [agi] Re: Defining AGI
Domain effectiveness (a.k.a. intelligence) is predicated upon having an effective internal model of that domain. Language production is the extraction and packaging of applicable parts of the internal model for transmission to others. Conversely, language understanding is for the reception (and integration) of model portions developed by others (i.e. learning from a teacher). The better your internal models, the more effective/intelligent you are. BUT! This also holds true for language! Concrete unadorned statements convey a lot less information than statements loaded with adjectives, adverbs, or even more markedly analogies (or innuendos or . . . ). A child cannot pick up the same amount of information from a sentence that they think that they understand (and do understand to some degree) that an adult can. Language is a knowledge domain like any other and high intelligences can use it far more effectively than lower intelligences. ** Or, in other words, I am disagreeing with the statement that the process itself needs not much intelligence. Saying that the understanding of language itself is simple is like saying that chess is simple because you understand the rules of the game. Godel's Incompleteness Theorem can be used to show that there is no upper bound on the complexity of language and the intelligence necessary to pack and extract meaning/knowledge into/from language. Language is *NOT* just a top-level communications protocol because it is not fully-specified and because it is tremendously context-dependent (not to mention entirely Godellian). These two reasons are why it *IS* inextricably tied into intelligence. I *might* agree that the concrete language of lower primates and young children is separate from intelligence, but there is far more going on in adult language than a simple communications protocol. E-mail programs are simply point-to-point repeaters of language (NOT meaning!) Intelligences generally don't exactly repeat language but *try* to repeat meaning. The game of telephone is a tremendous example of why language *IS* tied to intelligence (or look at the results of translating simple phrases into another language and back -- The drink is strong but the meat is rotten). Translating language to and from meaning (i.e. your domain model) is the essence of intelligence. How simple is the understanding of the above? How much are you having to fight to relate it to your internal model (assuming that it's even compatible :-)? I don't believe that intelligence is inherent upon language EXCEPT that language is necessary to convey knowledge/meaning (in order to build intelligence in a reasonable timeframe) and that language is influenced by and influences intelligence since it is basically the core of the critical meta-domains of teaching, learning, discovery, and alteration of your internal model (the effectiveness of which *IS* intelligence). Future AGI and humans will undoubtedly not only have a much richer language but also a much richer repertoire of second-order (and higher) features expressed via language. ** Or, in other words, I am strongly disagreeing that intelligence is separated from language understanding. I believe that language understanding is the necessary tool that intelligence is built with since it is what puts the *contents* of intelligence (i.e. the domain model) into intelligence . Trying to build an intelligence without language understanding is like trying to build it with just machine language or by using only observable data points rather than trying to build those things into more complex entities like third-, fourth-, and fifth-generation programming languages instead of machine language and/or knowledge instead of just data points. BTW -- Please note, however, that the above does not imply that I believe that NLU is the place to start in developing AGI. Quite the contrary -- NLU rests upon such a large domain model that I believe that it is counter-productive to start there. I believe that we need to star with limited domains and learn about language, internal models, and grounding without brittleness in tractable domains before attempting to extend that knowledge to larger domains. - Original Message - From: David Hart To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 5:30 AM Subject: Re: AW: [agi] Re: Defining AGI An excellent post, thanks! IMO, it raises the bar for discussion of language and AGI, and should be carefully considered by the authors of future posts on the topic of language and AGI. If the AGI list were a forum, Matthias's post should be pinned! -dave On Sun, Oct 19, 2008 at 6:58 PM, Dr. Matthias Heger [EMAIL PROTECTED] wrote: The process of outwardly expressing meaning may be fundamental to any social intelligence but the process itself needs not much intelligence. Every email program can receive meaning, store meaning and
Re: [agi] Words vs Concepts [ex Defining AGI]
There is no creation of new patterns and there is no intelligent algorithm which manipulates patterns. It is just translating, sending, receiving and retranslating. This is what I disagree entirely with. If nothing else, humans are constantly building and updating their mental model of what other people believe and how they communicate it. Only in routine, pre-negotiated conversations can language be entirely devoid of learning. Unless a conversation is entirely concrete and based upon something like shared physical experiences, it can't be any other way. You're only paying attention to the absolutely simplest things that language does (i.e. the tip of the iceberg). - Original Message - From: Dr. Matthias Heger [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 10:31 AM Subject: AW: [agi] Words vs Concepts [ex Defining AGI] For the discussion of the subject the details of the pattern representation are not important at all. It is sufficient if you agree that a spoken sentence represent a certain set of patterns which are translated into the sentence. The receiving agent retranslates the sentence and matches the content with its model by activating similar patterns. The activation of patterns is extremely fast and happens in real time. The brain even predicts patterns if it just hears the first syllable of a word: http://www.rochester.edu/news/show.php?id=3244 There is no creation of new patterns and there is no intelligent algorithm which manipulates patterns. It is just translating, sending, receiving and retranslating. From the ambiguities of natural language you obtain some hints about the structure of the patterns. But you cannot even expect to obtain all detail of these patterns by understanding the process of language understanding. There will be probably many details within these patterns which are only necessary for internal calculations. These details will be not visible from the linguistic point of view. Just think about communicating computers and you will know what I mean. - Matthias Mike Tintner [mailto:[EMAIL PROTECTED] wrote: Matthias, You seem - correct me - to be going a long way round saying that words are different from concepts - they're just sound-and-letter labels for concepts, which have a very different form. And the processing of words/language is distinct from and relatively simple compared to the processing of the underlying concepts. So take THE CAT SAT ON THE MAT or THE MIND HAS ONLY CERTAIN PARTS WHICH ARE SENTIENT or THE US IS THE HOME OF THE FINANCIAL CRISIS the words c-a-t or m-i-n-d or U-S or f-i-n-a-n-c-i-a-l c-r-i-s-i-s are distinct from the underlying concepts. The question is: What form do those concepts take? And what is happening in our minds (and what has to happen in any mind) when we process those concepts? You talk of patterns. What patterns, do you think, form the concept of mind that are engaged in thinking about sentence 2? Do you think that concepts like mind or the US might involve something much more complex still? Models? Or is that still way too simple? Spaces? Equally, of course, we can say that each *sentence* above is not just a verbal composition but a conceptual composition - and the question then is what form does such a composition take? Do sentences form, say, a pattern of patterns, or something like a picture? Or a blending of spaces ? Or are concepts like *money*? YOU CAN BUY A LOT WITH A MILLION DOLLARS Does every concept function somewhat like money, e.g. a million dollars - something that we know can be cashed in, in an infinite variety of ways, but that we may not have to start cashing in, (when processing), unless really called for - or only cash in so far? P.S. BTW this is the sort of psycho-philosophical discussion that I would see as central to AGI, but that most of you don't want to talk about? Matthias: What the computer makes with the data it receives depends on the information of the transferred data, its internal algorithms and its internal data. This is the same with humans and natural language. Language understanding would be useful to teach the AGI with existing knowledge already represented in natural language. But natural language understanding suffers from the problem of ambiguities. These ambiguities can be solved by having similar knowledge as humans have. But then you have a recursive problem because first there has to be solved the problem to obtain this knowledge. Nature solves this problem with embodiment. Different people make similar experiences since the laws of nature do not depend on space and time. Therefore we all can imagine a dog which is angry. Since we have experienced angry dogs but we haven't experienced angry trees we can resolve the linguistic ambiguity of my former example and answer the question: Who was angry? The way to obtain knowledge with embodiment is hard and long even in virtual worlds. If the
Re: [agi] Words vs Concepts [ex Defining AGI]
These details will be not visible from the linguistic point of view. Just think about communicating computers and you will know what I mean. Read Pinker's The Stuff of Thought. Actually, a lot of these details *are* visible from a linguistic point of view. - Original Message - From: Dr. Matthias Heger [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 10:31 AM Subject: AW: [agi] Words vs Concepts [ex Defining AGI] For the discussion of the subject the details of the pattern representation are not important at all. It is sufficient if you agree that a spoken sentence represent a certain set of patterns which are translated into the sentence. The receiving agent retranslates the sentence and matches the content with its model by activating similar patterns. The activation of patterns is extremely fast and happens in real time. The brain even predicts patterns if it just hears the first syllable of a word: http://www.rochester.edu/news/show.php?id=3244 There is no creation of new patterns and there is no intelligent algorithm which manipulates patterns. It is just translating, sending, receiving and retranslating. From the ambiguities of natural language you obtain some hints about the structure of the patterns. But you cannot even expect to obtain all detail of these patterns by understanding the process of language understanding. There will be probably many details within these patterns which are only necessary for internal calculations. These details will be not visible from the linguistic point of view. Just think about communicating computers and you will know what I mean. - Matthias Mike Tintner [mailto:[EMAIL PROTECTED] wrote: Matthias, You seem - correct me - to be going a long way round saying that words are different from concepts - they're just sound-and-letter labels for concepts, which have a very different form. And the processing of words/language is distinct from and relatively simple compared to the processing of the underlying concepts. So take THE CAT SAT ON THE MAT or THE MIND HAS ONLY CERTAIN PARTS WHICH ARE SENTIENT or THE US IS THE HOME OF THE FINANCIAL CRISIS the words c-a-t or m-i-n-d or U-S or f-i-n-a-n-c-i-a-l c-r-i-s-i-s are distinct from the underlying concepts. The question is: What form do those concepts take? And what is happening in our minds (and what has to happen in any mind) when we process those concepts? You talk of patterns. What patterns, do you think, form the concept of mind that are engaged in thinking about sentence 2? Do you think that concepts like mind or the US might involve something much more complex still? Models? Or is that still way too simple? Spaces? Equally, of course, we can say that each *sentence* above is not just a verbal composition but a conceptual composition - and the question then is what form does such a composition take? Do sentences form, say, a pattern of patterns, or something like a picture? Or a blending of spaces ? Or are concepts like *money*? YOU CAN BUY A LOT WITH A MILLION DOLLARS Does every concept function somewhat like money, e.g. a million dollars - something that we know can be cashed in, in an infinite variety of ways, but that we may not have to start cashing in, (when processing), unless really called for - or only cash in so far? P.S. BTW this is the sort of psycho-philosophical discussion that I would see as central to AGI, but that most of you don't want to talk about? Matthias: What the computer makes with the data it receives depends on the information of the transferred data, its internal algorithms and its internal data. This is the same with humans and natural language. Language understanding would be useful to teach the AGI with existing knowledge already represented in natural language. But natural language understanding suffers from the problem of ambiguities. These ambiguities can be solved by having similar knowledge as humans have. But then you have a recursive problem because first there has to be solved the problem to obtain this knowledge. Nature solves this problem with embodiment. Different people make similar experiences since the laws of nature do not depend on space and time. Therefore we all can imagine a dog which is angry. Since we have experienced angry dogs but we haven't experienced angry trees we can resolve the linguistic ambiguity of my former example and answer the question: Who was angry? The way to obtain knowledge with embodiment is hard and long even in virtual worlds. If the AGI shall understand natural language it would be necessary that it makes similar experiences as humans make in the real world. But this would need a very very sophisticated and rich virtual world. At least, there have to be angry dogs in the virtual world ;-) As I have already said I do not think the relation between utility of this approach and the costs would be positive for first AGI.
AW: [agi] Words vs Concepts [ex Defining AGI]
The process of changing the internal model does not belong to language understanding. Language understanding ends if the matching process is finished. Language understanding can be strictly separated conceptually from creation and manipulation of patterns as you can separate the process of communication with the process of manipulating the database in a computer. You can see it differently but then everything is only a discussion about definitions. - Matthias Mark Waser [mailto:[EMAIL PROTECTED] wrote Gesendet: Sonntag, 19. Oktober 2008 19:00 An: agi@v2.listbox.com Betreff: Re: [agi] Words vs Concepts [ex Defining AGI] There is no creation of new patterns and there is no intelligent algorithm which manipulates patterns. It is just translating, sending, receiving and retranslating. This is what I disagree entirely with. If nothing else, humans are constantly building and updating their mental model of what other people believe and how they communicate it. Only in routine, pre-negotiated conversations can language be entirely devoid of learning. Unless a conversation is entirely concrete and based upon something like shared physical experiences, it can't be any other way. You're only paying attention to the absolutely simplest things that language does (i.e. the tip of the iceberg). --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
AW: [agi] Words vs Concepts [ex Defining AGI]
If there are some details of the internal structure of patterns visible then this is no proof at all that there are not also details of the structure which are completely hidden from the linguistic point of view. Since in many communicating technical systems there are so much details which are not transferred I would bet that this is also the case in humans. As long as we have no proof this remains an open question. An AGI which may have internal features for its patterns would have less restrictions and is thus far easier to build. - Matthias. Mark Waser [mailto:[EMAIL PROTECTED] wrote Read Pinker's The Stuff of Thought. Actually, a lot of these details *are* visible from a linguistic point of view. --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
AW: AW: [agi] Re: Defining AGI
The process of translating patterns into language should be easier than the process of creating patterns or manipulating patterns. Therefore I say that language understanding is easy. When you say that language is not fully specified then you probably imagine an AGI which learns language. This is a complete different thing. Learning language is difficult as I already have mentioned. Language cannot be translated into meaning. Meaning is a mapping from a linguistic string to patterns. Email programs are not just point to point repeaters. They receive data in a certain communication protocol. They translate these data into an internal representation and store the data. And they can translate their internal data into a linguistic representation to send the data to another email client. This process of communication is conceptually the same as we can observe it with humans. The word meaning was bad chosen from me. But brains do not transfer meaning as well. They also just transfer data. Meaning is a mapping. You *believe* that language cannot be separated from intelligence. I don't and I have described a model which has a strict separation. We both have no proof. - Matthias Mark Waser [mailto:[EMAIL PROTECTED] wrote BUT! This also holds true for language! Concrete unadorned statements convey a lot less information than statements loaded with adjectives, adverbs, or even more markedly analogies (or innuendos or . . . ). A child cannot pick up the same amount of information from a sentence that they think that they understand (and do understand to some degree) that an adult can. Language is a knowledge domain like any other and high intelligences can use it far more effectively than lower intelligences. ** Or, in other words, I am disagreeing with the statement that the process itself needs not much intelligence. Saying that the understanding of language itself is simple is like saying that chess is simple because you understand the rules of the game. Godel's Incompleteness Theorem can be used to show that there is no upper bound on the complexity of language and the intelligence necessary to pack and extract meaning/knowledge into/from language. Language is *NOT* just a top-level communications protocol because it is not fully-specified and because it is tremendously context-dependent (not to mention entirely Godellian). These two reasons are why it *IS* inextricably tied into intelligence. I *might* agree that the concrete language of lower primates and young children is separate from intelligence, but there is far more going on in adult language than a simple communications protocol. E-mail programs are simply point-to-point repeaters of language (NOT meaning!) Intelligences generally don't exactly repeat language but *try* to repeat meaning. The game of telephone is a tremendous example of why language *IS* tied to intelligence (or look at the results of translating simple phrases into another language and back -- The drink is strong but the meat is rotten). Translating language to and from meaning (i.e. your domain model) is the essence of intelligence. How simple is the understanding of the above? How much are you having to fight to relate it to your internal model (assuming that it's even compatible :-)? I don't believe that intelligence is inherent upon language EXCEPT that language is necessary to convey knowledge/meaning (in order to build intelligence in a reasonable timeframe) and that language is influenced by and influences intelligence since it is basically the core of the critical meta-domains of teaching, learning, discovery, and alteration of your internal model (the effectiveness of which *IS* intelligence). Future AGI and humans will undoubtedly not only have a much richer language but also a much richer repertoire of second-order (and higher) features expressed via language. ** Or, in other words, I am strongly disagreeing that intelligence is separated from language understanding. I believe that language understanding is the necessary tool that intelligence is built with since it is what puts the *contents* of intelligence (i.e. the domain model) into intelligence . Trying to build an intelligence without language understanding is like trying to build it with just machine language or by using only observable data points rather than trying to build those things into more complex entities like third-, fourth-, and fifth-generation programming languages instead of machine language and/or knowledge instead of just data points. BTW -- Please note, however, that the above does not imply that I believe that NLU is the place to start in developing AGI. Quite the contrary -- NLU rests upon such a large domain model that I believe that it is counter-productive to start there. I believe that we need to star with limited domains and learn about language, internal models, and grounding without brittleness in
Re: RSI without input (was Re: [agi] Updated AGI proposal (CMR v2.1))
--- On Sun, 10/19/08, Samantha Atkins [EMAIL PROTECTED] wrote: Matt Mahoney wrote: There is currently a global brain (the world economy) with an IQ of around 10^10, and approaching 10^12. Oh man. It is so tempting in today's economic morass to point out the obvious stupidity of this purported super-super-genius. Why would you assign such an astronomical intelligence to the economy? Without the economy, or the language and culture needed to support it, you would be foraging for food and sleeping in the woods. You would not know that you could grow crops by planting seeds, or that you could make a spear out of sticks and rocks and use it for hunting. There is a 99.9% chance that you would starve because the primitive earth could only support a few million humans, not a few billions. I realize it makes no sense to talk of an IQ of 10^10 when current tests only go to about 200. But by any measure of goal achievement, such as dollars earned or number of humans that can be supported, the global brain has enormous intelligence. It is a known fact that groups of humans collectively make more accurate predictions than their members, e.g. prediction markets. http://en.wikipedia.org/wiki/Prediction_market Such markets would not work if the members did not individually think that they were smarter than the group (i.e. disagree). You may think you could run the government better than current leadership, but it is a fact that people are better off (as measured by GDP and migration) in democracies than dictatorships. Group decision making is also widely used in machine learning, e.g. the PAQ compression programs. How much of the advancement of humanity is the result of a relatively few exceptionally bright minds rather than the billions of lesser intelligences? Very little, because agents at any intelligence level cannot detect higher intelligence. Socrates was executed. Galileo was arrested. Even today, there is a span of decades between pioneering scientific work and its recognition with a Nobel prize. So I don't expect anyone to recognize the intelligence of the economy. But your ability to read this email depends more on circuit board assemblers in Malaysia than you are willing to give the world credit for. -- Matt Mahoney, [EMAIL PROTECTED] --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] Words vs Concepts [ex Defining AGI]
The process of changing the internal model does not belong to language understanding. Language understanding ends if the matching process is finished. What if the matching process is not finished? This is overly simplistic for several reasons since you're apparently assuming that the matching process is crisp, unambiguous, and irreversible (and ask Stephen Reed how well that works for TexAI). It *must* be remembered that the internal model for natural language includes such critically entwined and constantly changing information as what this particular conversation is about, what the speaker knows, and what the speakers motivations are. The meaning of sentences can change tremendously based upon the currently held beliefs about these questions. Suddenly realizing that the speaker is being sarcastic generally reverses the meaning of statements. Suddenly realizing that the speaker is using an analogy can open up tremendous vistas for interpretation and analysis. Look at all the problems that people have parsing sentences. Language understanding can be strictly separated conceptually from creation and manipulation of patterns as you can separate the process of communication with the process of manipulating the database in a computer. The reason why you can separate the process of communication with the process of manipulating data in a computer is because *data* is crisp and unambiguous. It is concrete and completely specified as I suggested in my initial e-mail. The model is entirely known and the communication process is entirely specified. None of these things are true of unstructured knowledge. Language understanding emphatically does not meet these requirements so your analogy doesn't hold. You can see it differently but then everything is only a discussion about definitions. No, and claiming that everything is just a discussion about definitions is a strawman. Your analogies are not accurate and your model is incomplete. You are focusing only on the tip of the iceberg (concrete language as spoken by a two-year-old) and missing the essence of NLP. - Original Message - From: Dr. Matthias Heger [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 1:42 PM Subject: AW: [agi] Words vs Concepts [ex Defining AGI] The process of changing the internal model does not belong to language understanding. Language understanding ends if the matching process is finished. Language understanding can be strictly separated conceptually from creation and manipulation of patterns as you can separate the process of communication with the process of manipulating the database in a computer. You can see it differently but then everything is only a discussion about definitions. - Matthias Mark Waser [mailto:[EMAIL PROTECTED] wrote Gesendet: Sonntag, 19. Oktober 2008 19:00 An: agi@v2.listbox.com Betreff: Re: [agi] Words vs Concepts [ex Defining AGI] There is no creation of new patterns and there is no intelligent algorithm which manipulates patterns. It is just translating, sending, receiving and retranslating. This is what I disagree entirely with. If nothing else, humans are constantly building and updating their mental model of what other people believe and how they communicate it. Only in routine, pre-negotiated conversations can language be entirely devoid of learning. Unless a conversation is entirely concrete and based upon something like shared physical experiences, it can't be any other way. You're only paying attention to the absolutely simplest things that language does (i.e. the tip of the iceberg). --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
AW: AW: AW: [agi] Re: Defining AGI
Terren wrote: Isn't the *learning* of language the entire point? If you don't have an answer for how an AI learns language, you haven't solved anything. The understanding of language only seems simple from the point of view of a fluent speaker. Fluency however should not be confused with a lack of intellectual effort - rather, it's a state in which the effort involved is automatic and beyond awareness. I don't think that learning of language is the entire point. If I have only learned language I still cannot create anything. A human who can understand language is by far still no good scientist. Intelligence means the ability to solve problems. Which problems can a system solve if it can nothing else than language understanding? Einstein had to express his (non-linguistic) internal insights in natural language and in mathematical language. In both modalities he had to use his intelligence to make the translation from his mental models. The point is that someone else could understand Einstein even if he haven't had the same intelligence. This is a proof that understanding AI1 does not necessarily imply to have the intelligence of AI1. Deaf people speak in sign language, which is only different from spoken language in superficial ways. This does not tell us much about language that we didn't already know. But it is a proof that *natural* language understanding is not necessary for human-level intelligence. It is surely true that much/most of our cognitive processing is not at all linguistic, and that there is much that happens beyond our awareness. However, language is a necessary tool, for humans at least, to obtain a competent conceptual framework, even if that framework ultimately transcends the linguistic dynamics that helped develop it. Without language it is hard to see how humans could develop self-reflectivity. I have already outlined the process of self-reflectivity: Internal patterns are translated into language. This is routed to the brain's own input regions. You *hear* your own thoughts and have the illusion that you think linguistically. If you can speak two languages then you can make an easy test: Try to think in the foreign language. It works. If language would be inherently involved in the process of thoughts then thinking alternatively in two languages would cost many resources of the brain. In fact you need just use the other module for language translation. This is a big hint that language and thoughts do not have much in common. -Matthias --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] Words vs Concepts [ex Defining AGI]
If there are some details of the internal structure of patterns visible then this is no proof at all that there are not also details of the structure which are completely hidden from the linguistic point of view. True, but visible patterns offer clues for interpretation and analysis. The more that is visible and clear, the less that is ambiguous and needs to be guessed at. This is where your analogy splitting computer communications and data updates is accurate because the internal structures have been communicated and are shared to the nth degree. Since in many communicating technical systems there are so much details which are not transferred I would bet that this is also the case in humans. Details that don't need to be transferred are those which are either known by or unnecessary to the recipient. The former is a guess (unless the details were transmitted previously) and the latter is an assumption based upon partial knowledge of the recipient. In a perfect, infinite world, details could and should always be transferred. In the real world, time and computational constraints means that trade-offs need to occur. This is where the essence of intelligence comes into play -- determining which of the trade-offs to take to get optimal perfomance (a.k.a. domain competence) As long as we have no proof this remains an open question. What remains an open question? Obviously there are details which can be teased out by behavior and details that can't be easily teased out because we have insufficient data to do so. This is like any other scientific examination of any other complex phenomenon. An AGI which may have internal features for its patterns would have less restrictions and is thus far easier to build. Sorry, but I can't interpret this. An AGI without internal features and regularities is an oxymoron and completely nonsensical. What are you trying to convey here? --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
AW: [agi] Words vs Concepts [ex Defining AGI]
Mark Waser wrote What if the matching process is not finished? This is overly simplistic for several reasons since you're apparently assuming that the matching process is crisp, unambiguous, and irreversible (and ask Stephen Reed how well that works for TexAI). I do not assume this. Why should I? It *must* be remembered that the internal model for natural language includes such critically entwined and constantly changing information as what this particular conversation is about, what the speaker knows, and what the speakers motivations are. The meaning of sentences can change tremendously based upon the currently held beliefs about these questions. Suddenly realizing that the speaker is being sarcastic generally reverses the meaning of statements. Suddenly realizing that the speaker is using an analogy can open up tremendous vistas for interpretation and analysis. Look at all the problems that people have parsing sentences. If I suddenly realize that the speaker is sarcastic than I change my mappings from linguistic entities to pattern entities. Where is the problem? The reason why you can separate the process of communication with the process of manipulating data in a computer is because *data* is crisp and unambiguous. It is concrete and completely specified as I suggested in my initial e-mail. The model is entirely known and the communication process is entirely specified. None of these things are true of unstructured knowledge. You have given no reason why the separation of the process of communication with the process of manipulating data can only be separated if the knowledge is structured. In fact there is no reason. Language understanding emphatically does not meet these requirements so your analogy doesn't hold. There are no special requirements. - Matthias --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
AW: [agi] Words vs Concepts [ex Defining AGI]
We can assume that the speaking human itself is not aware about every details of its patterns. At least these details would be probably hidden from communication. -Matthias Mark Waser wrote Details that don't need to be transferred are those which are either known by or unnecessary to the recipient. The former is a guess (unless the details were transmitted previously) and the latter is an assumption based upon partial knowledge of the recipient. In a perfect, infinite world, details could and should always be transferred. In the real world, time and computational constraints means that trade-offs need to occur. This is where the essence of intelligence comes into play -- determining which of the trade-offs to take to get optimal perfomance (a.k.a. domain competence) --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: [agi] Re: Defining AGI
The process of translating patterns into language should be easier than the process of creating patterns or manipulating patterns. How is translating patterns into language different from manipulating patterns? It seems to me that they are *exactly* the same thing. How do you believe that they differ? Therefore I say that language understanding is easy. Do you really believe that if A is easier than B then that makes A easy? How about if A is leaping a tall building in a single bound and B is jumping to the moon? When you say that language is not fully specified then you probably imagine an AGI which learns language. Do you believe that language is fully specified? That we can program English into an AGI by hand? Yes, I imagine that an AGI must have some process for learning language because language is necessary for learning knowledge and knowledge is necessary for intelligence. What part of that do you disagree with? Please be specific. This is a complete different thing. Learning language is difficult as I already have mentioned. And this is where we are not communicating. Since language is not fully specified, then the participants in many conversations are *constantly* creating and learning language as a part of the process of communication. This is where Gödel's incompleteness comes in. To be a General Intelligence, you must be able to extend beyond what is currently known and specified into new domains. Any time that we are teaching or learning (i.e. modifying our model of the world), we are also necessarily extending our models of each other and language. The computer database analogy you are basing your entire argument upon does not have the necessary features/complexity to be an accurate or useful analogy. Email programs are not just point to point repeaters. They receive data in a certain communication protocol. They translate these data into an internal representation and store the data. And they can translate their internal data into a linguistic representation to send the data to another email client. This process of communication is conceptually the same as we can observe it with humans. Again, I disagree. You added internal details but the end result after the details are hidden is that e-mail programs are just point-to-point repeaters. That is why I used the examples (the telephone game and round-trip (mis)translations) that I did which you did not address. The word meaning was bad chosen from me. But brains do not transfer meaning as well. They also just transfer data. Meaning is a mapping. As I said, brains *try* to transfer meaning (though they must do it via the transfer of data). If you don't believe that brains try (and most frequently succeed) at transferring meaning they we should just agree to disagree. You *believe* that language cannot be separated from intelligence. I don't and I have described a model which has a strict separation. We both have no proof. Three points. 1. My statement was that intelligence can't be built without language/communication. That is entirely different from the fact that they can't be separated. I also gave reasoning why this was the case that you haven't addressed. 2. Your model has serious flaws that you have not answered. You are relying upon an analogy that has points that you have not shown that you are able to defend. Until you do so, this invalidates your model. 3. You have not provided a disproof or counter-example to what I am saying. I have clearly specified where your analogy comes up short and other inaccuracies in your statements while you have not done so for any of mine (other than of the tis too, tis not variety). I have had the courtesy to directly address your points with clear counter-examples. Please return the favor and do not simply drop my examples without replying to them and revert back to global statements. Global statements are great for an initial exposition but eventually you have to get down to the details and work out the nitty-gritty. Thanks. Mark - Original Message - From: Dr. Matthias Heger To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 2:19 PM Subject: AW: AW: [agi] Re: Defining AGI The process of translating patterns into language should be easier than the process of creating patterns or manipulating patterns. Therefore I say that language understanding is easy. When you say that language is not fully specified then you probably imagine an AGI which learns language. This is a complete different thing. Learning language is difficult as I already have mentioned. Language cannot be translated into meaning. Meaning is a mapping from a linguistic string to patterns. Email programs are not just point to point repeaters. They receive data in a certain communication protocol. They translate these data
AW: [agi] Re: Meaning, communication and understanding
The language model does not need interaction with the environment when the language model is already complete which is possible for formal languages but nearly impossible for natural language. That is the reason why formal language need much less cost. If the language must be learned then things are completely different and you are right that the interaction with the environment is necessary to learn L. But in any case there is a complete distinction between D and L. The brain never sends entities of D to its output region but it sends entities of L. Therefore there must be a strict separation between language model and D. - Matthias Vladimir Nesov wrote I think that this model is overly simplistic, overemphasizing an artificial divide between domains within AI's cognition (L and D), and externalizing communication domain from the core of AI. Both world model and language model support interaction with environment, there is no clear cognitive distinction between them. As a given, interaction happens at the narrow I/O interface, and anything else is a design decision for a specific AI (even invariability of I/O is, a simplifying assumption that complicates semantics of time and more radical self-improvement). Sufficiently flexible cognitive algorithm should be able to integrate facts about any domain, becoming able to generate appropriate behavior in corresponding contexts. --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: AW: [agi] Re: Defining AGI
I don't think that learning of language is the entire point. If I have only learned language I still cannot create anything. A human who can understand language is by far still no good scientist. Intelligence means the ability to solve problems. Which problems can a system solve if it can nothing else than language understanding? Many or most people on this list believe that learning language is an AGI-complete task. What this means is that the skills necessary for learning a language are necessary and sufficient for learning any other task. It is not that language understanding gives general intelligence capabilities, but that the pre-requisites for language understanding are general intelligence (or, that language understanding is isomorphic to general intelligence in the same fashion that all NP-complete problems are isomorphic). Thus, the argument actually is that a system that can do nothing else than language understanding is an oxymoron. *Any* human who can understand language beyond a certain point (say, that of a slightly sub-average human IQ) can easily be taught to be a good scientist if they are willing to play along. Science is a rote process that can be learned and executed by anyone -- as long as their beliefs and biases don't get in the way. Deaf people speak in sign language, which is only different from spoken language in superficial ways. This does not tell us much about language that we didn't already know. But it is a proof that *natural* language understanding is not necessary for human-level intelligence. This is a bit of disingenuous side-track that I feel that I must address. When people say natural language, the important features are extensibility and ambiguity. If you can handle one extensible and ambiguous language, you should have the capabilities to handle all of them. It's yet another definition of GI-complete. Just look at it as yet another example of dealing competently with ambiguous and incomplete data (which is, at root, all that intelligence is). If you can speak two languages then you can make an easy test: Try to think in the foreign language. It works. If language would be inherently involved in the process of thoughts then thinking alternatively in two languages would cost many resources of the brain. In fact you need just use the other module for language translation. This is a big hint that language and thoughts do not have much in common. One thought module, two translation modules -- except that all the translation modules really are is label appliers and grammar re-arrangers. The heavy lifting is all in the thought module. The problem is that you are claiming that language lies entirely in the translation modules while I'm arguing that a large percentage of it is in the thought module. The fact that the translation module has to go to the thought module for disambiguation and interpretation (and numerous other things) should make it quite clear that language is *not* simply translation. Further, if you read Pinker's book, you will find that languages have a lot more in common than you would expect if language truly were independent of and separate from thought (as you are claiming). Language is built on top of the thinking/cognitive architecture (not beside it and not independent of it) and could not exist without it. That is why language is AGI-complete. Language also gives an excellent window into many of the features of that cognitive architecture and determining what is necessary for language also determine what is in that cognitive architecture. Another excellent window is how humans perform moral judgments (try reading Marc Hauser -- either his numerous scientific papers or the excellent Moral Minds). Or, yet another, is examining the structure of human biases. - Original Message - From: Dr. Matthias Heger [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 2:52 PM Subject: AW: AW: AW: [agi] Re: Defining AGI Terren wrote: Isn't the *learning* of language the entire point? If you don't have an answer for how an AI learns language, you haven't solved anything. The understanding of language only seems simple from the point of view of a fluent speaker. Fluency however should not be confused with a lack of intellectual effort - rather, it's a state in which the effort involved is automatic and beyond awareness. I don't think that learning of language is the entire point. If I have only learned language I still cannot create anything. A human who can understand language is by far still no good scientist. Intelligence means the ability to solve problems. Which problems can a system solve if it can nothing else than language understanding? Einstein had to express his (non-linguistic) internal insights in natural language and in mathematical language. In both modalities he had to use his intelligence to make the translation from his
Re: [agi] Words vs Concepts [ex Defining AGI]
You have given no reason why the separation of the process of communication with the process of manipulating data can only be separated if the knowledge is structured. In fact there is no reason. How do you communicate something for which you have no established communications protocol? If you can answer that, you have solved the natural language problem. - Original Message - From: Dr. Matthias Heger [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 3:10 PM Subject: AW: [agi] Words vs Concepts [ex Defining AGI] Mark Waser wrote What if the matching process is not finished? This is overly simplistic for several reasons since you're apparently assuming that the matching process is crisp, unambiguous, and irreversible (and ask Stephen Reed how well that works for TexAI). I do not assume this. Why should I? It *must* be remembered that the internal model for natural language includes such critically entwined and constantly changing information as what this particular conversation is about, what the speaker knows, and what the speakers motivations are. The meaning of sentences can change tremendously based upon the currently held beliefs about these questions. Suddenly realizing that the speaker is being sarcastic generally reverses the meaning of statements. Suddenly realizing that the speaker is using an analogy can open up tremendous vistas for interpretation and analysis. Look at all the problems that people have parsing sentences. If I suddenly realize that the speaker is sarcastic than I change my mappings from linguistic entities to pattern entities. Where is the problem? The reason why you can separate the process of communication with the process of manipulating data in a computer is because *data* is crisp and unambiguous. It is concrete and completely specified as I suggested in my initial e-mail. The model is entirely known and the communication process is entirely specified. None of these things are true of unstructured knowledge. You have given no reason why the separation of the process of communication with the process of manipulating data can only be separated if the knowledge is structured. In fact there is no reason. Language understanding emphatically does not meet these requirements so your analogy doesn't hold. There are no special requirements. - Matthias --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] Words vs Concepts [ex Defining AGI]
We can assume that the speaking human itself is not aware about every details of its patterns. At least these details would be probably hidden from communication. Absolutely. We are not aware of most of our assumptions that are based in our common heritage, culture, and embodiment. But an external observer could easily notice them and tease out an awful lot of information about us by doing so. - Original Message - From: Dr. Matthias Heger [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 3:18 PM Subject: AW: [agi] Words vs Concepts [ex Defining AGI] We can assume that the speaking human itself is not aware about every details of its patterns. At least these details would be probably hidden from communication. -Matthias Mark Waser wrote Details that don't need to be transferred are those which are either known by or unnecessary to the recipient. The former is a guess (unless the details were transmitted previously) and the latter is an assumption based upon partial knowledge of the recipient. In a perfect, infinite world, details could and should always be transferred. In the real world, time and computational constraints means that trade-offs need to occur. This is where the essence of intelligence comes into play -- determining which of the trade-offs to take to get optimal perfomance (a.k.a. domain competence) --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] Re: Meaning, communication and understanding
The language model does not need interaction with the environment when the language model is already complete which is possible for formal languages but nearly impossible for natural language. That is the reason why formal language need much less cost. Yes! But the formal languages need to be efficiently extensible as well (and ambiguity plays a large part in extensibility which then leads to . . . . :-) If the language must be learned then things are completely different and you are right that the interaction with the environment is necessary to learn L. How do you go from a formal language to a competent description of a messy, ambiguous, data-deficient world? *That* is the natural language question. What happens if I say that language extensibility is exactly analogous to learning which is exactly analogous to internal model improvement? But in any case there is a complete distinction between D and L. The brain never sends entities of D to its output region but it sends entities of L. Therefore there must be a strict separation between language model and D. I disagree with a complete distinction between D and L. L is a very small fraction of D translated for transmission. However, instead of arguing that there must be a strict separation between language model and D, I would argue that the more similar the two could be (i.e. the less translation from D to L) the better. Analyzing L in that case could tell you more about D than you might think (which is what Pinker and Hauser argue). It's like looking at data to determine an underlying cause for a phenomenon. Even noticing what does and does not vary (and what covaries) tells you a lot about the underlying cause (D). - Original Message - From: Dr. Matthias Heger [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 3:50 PM Subject: AW: [agi] Re: Meaning, communication and understanding The language model does not need interaction with the environment when the language model is already complete which is possible for formal languages but nearly impossible for natural language. That is the reason why formal language need much less cost. If the language must be learned then things are completely different and you are right that the interaction with the environment is necessary to learn L. But in any case there is a complete distinction between D and L. The brain never sends entities of D to its output region but it sends entities of L. Therefore there must be a strict separation between language model and D. - Matthias Vladimir Nesov wrote I think that this model is overly simplistic, overemphasizing an artificial divide between domains within AI's cognition (L and D), and externalizing communication domain from the core of AI. Both world model and language model support interaction with environment, there is no clear cognitive distinction between them. As a given, interaction happens at the narrow I/O interface, and anything else is a design decision for a specific AI (even invariability of I/O is, a simplifying assumption that complicates semantics of time and more radical self-improvement). Sufficiently flexible cognitive algorithm should be able to integrate facts about any domain, becoming able to generate appropriate behavior in corresponding contexts. --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: AW: AW: [agi] Re: Defining AGI
Matthias wrote: I don't think that learning of language is the entire point. If I have only learned language I still cannot create anything. A human who can understand language is by far still no good scientist. Intelligence means the ability to solve problems. Which problems can a system solve if it can nothing else than language understanding? Language understanding requires a sophisticated conceptual framework complete with causal models, because, whatever meaning means, it must be captured somehow in an AI's internal models of the world. The Piraha tribe in the Amazon basin has a very primitive language compared to all modern languages - it has no past or future tenses, for example - and as a people they exhibit barely any of the hallmarks of abstract reasoning that are so common to the rest of humanity, such as story-telling, artwork, religion... see http://en.wikipedia.org/wiki/Pirah%C3%A3_people. How do you explain that? Einstein had to express his (non-linguistic) internal insights in natural language and in mathematical language. In both modalities he had to use his intelligence to make the translation from his mental models. The point is that someone else could understand Einstein even if he haven't had the same intelligence. This is a proof that understanding AI1 does not necessarily imply to have the intelligence of AI1. I'm saying that if an AI understands speaks natural language, you've solved AGI - your Nobel will be arriving soon. The difference between AI1 that understands Einstein, and any AI currently in existence, is much greater then the difference between AI1 and Einstein. Deaf people speak in sign language, which is only different from spoken language in superficial ways. This does not tell us much about language that we didn't already know. But it is a proof that *natural* language understanding is not necessary for human-level intelligence. Sorry, I don't see that, can you explain the proof? Are you saying that sign language isn't natural language? That would be patently false. (see http://crl.ucsd.edu/signlanguage/) I have already outlined the process of self-reflectivity: Internal patterns are translated into language. So you're agreeing that language is necessary for self-reflectivity. In your models, then, self-reflectivity is not important to AGI, since you say AGI can be realized without language, correct? This is routed to the brain's own input regions. You *hear* your own thoughts and have the illusion that you think linguistically. If you can speak two languages then you can make an easy test: Try to think in the foreign language. It works. If language would be inherently involved in the process of thoughts then thinking alternatively in two languages would cost many resources of the brain. In fact you need just use the other module for language translation. This is a big hint that language and thoughts do not have much in common. -Matthias I'm not saying that language is inherently involved in thinking, but it is crucial for the development of *sophisticated* causal models of the world - the kind of models that can support self-reflectivity. Word-concepts form the basis of abstract symbol manipulation. That gets the ball rolling for humans, but the conceptual framework that emerges is not necessarily tied to linguistics, especially as humans get feedback from the world in ways that are not linguistic (scientific experimentation/tinkering, studying math, art, music, etc). Terren __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] constructivist issues
--- On Sat, 10/18/08, Abram Demski [EMAIL PROTECTED] wrote: No, I do not claim that computer theorem-provers cannot prove Goedel's Theorem. It has been done. The objection applies specifically to AIXI-- AIXI cannot prove goedel's theorem. Yes it can. It just can't understand its own proof in the sense of Tarski's undefinability theorem. Construct a predictive AIXI environment as follows: the environment output symbol does not depend on anything the agent does. However, the agent receives a reward when its output symbol matches the next symbol input from the environment. Thus, the environment can be modeled as a string that the agent has the goal of compressing. Now encode in the environment a series of theorems followed by their proofs. Since proofs can be mechanically checked, and therefore found given enough time (if the proof exists), then the optimal strategy for the agent, according to AIXI is to guess that the environment receives as input a series of theorems and that the environment then proves them and outputs the proof. AIXI then replicates its guess, thus correctly predicting the proofs and maximizing its reward. To prove Goedel's theorem, we simply encode it into the environment after a series of other theorems and their proofs. -- Matt Mahoney, [EMAIL PROTECTED] --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] constructivist issues
But, either you're just wrong or I don't understand your wording ... of course, AIXI **can** reason about uncomputable entities. If you showed AIXI the axioms of, say, ZF set theory (including the Axiom of Choice), and reinforced it for correctly proving theorems about uncomputable entities as defined in ZF, then after enough reinforcement signals it could learn to prove these theorems. ben g On Sun, Oct 19, 2008 at 10:42 AM, Abram Demski [EMAIL PROTECTED]wrote: Ben, I don't know what sounded almost confused, but anyway it is apparent that I didn't make my position clear. I am not saying we can manipulate these things directly via exotic (non)computing. First, I am very specifically saying that AIXI-style AI (meaning, any AI that approaches AIXI as resources increase) cannot reason about uncomputable entities. This is because AIXI entertains only computable models. Second, I am suggesting a broader problem that will apply to a wide class of formulations of idealized intelligence such as AIXI: if their internal logic obeys a particular set of assumptions, it will become prone to Tarski's Undefinability Theorem. Therefore, we humans will be able to point out a particular class of concepts that it cannot reason about; specifically, the very concepts used in describing the ideal intelligence in the first place. One reasonable way of avoiding the humans are magic explanation of this (or humans use quantum gravity computing, etc) is to say that, OK, humans really are an approximation of an ideal intelligence obeying those assumptions. Therefore, we cannot understand the math needed to define our own intelligence. Therefore, we can't engineer human-level AGI. I don't like this conclusion! I want a different way out. I'm not sure the guru explanation is enough... who was the Guru for Humankind? Thanks, --Abram On Sun, Oct 19, 2008 at 5:39 AM, Ben Goertzel [EMAIL PROTECTED] wrote: Abram, I find it more useful to think in terms of Chaitin's reformulation of Godel's Theorem: http://www.cs.auckland.ac.nz/~chaitin/sciamer.htmlhttp://www.cs.auckland.ac.nz/%7Echaitin/sciamer.html Given any computer program with algorithmic information capacity less than K, it cannot prove theorems whose algorithmic information content is greater than K. Put simply, there are some things our brains are not big enough to prove true or false This is true for quantum computers just as it's true for classical computers. Penrose hypothesized it would NOT hold for quantum gravity computers, but IMO this is a fairly impotent hypothesis because quantum gravity computers don't exist (even theoretically, I mean: since there is no unified quantum gravity theory yet). Penrose assumes that humans don't have this sort of limitation, but I'm not sure why. On the other hand, this limitation can be overcome somewhat if you allow the program P to interact with the external world in a way that lets it be modified into P1 such that P1 is not computable by P. In this case P needs to have a guru (or should I say an oracle ;-) that it trusts to modify itself in ways it can't understand, or else to be a gambler-type... You seem almost confused when you say that an AI can't reason about uncomputable entities. Of course it can. An AI can manipulate math symbols in a certain formal system, and then associate these symbols with the words uncomputable entities, and with its own self ... or us. This is what we do. An AI program can't actually manipulate the uncomputable entities directly , but what makes you think *we* can, either? -- Ben G --- 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 -- Ben Goertzel, PhD CEO, Novamente LLC and Biomind LLC Director of Research, SIAI [EMAIL PROTECTED] Nothing will ever be attempted if all possible objections must be first overcome - Dr Samuel Johnson --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
AW: AW: AW: [agi] Re: Defining AGI
Marc Walser wrote: *Any* human who can understand language beyond a certain point (say, that of a slightly sub-average human IQ) can easily be taught to be a good scientist if they are willing to play along. Science is a rote process that can be learned and executed by anyone -- as long as their beliefs and biases don't get in the way. This is just an opinion and I strongly disagree with your opinion. Obviously you overestimate language understanding a lot. This is a bit of disingenuous side-track that I feel that I must address. When people say natural language, the important features are extensibility and ambiguity. If you can handle one extensible and ambiguous language, you should have the capabilities to handle all of them. It's yet another definition of GI-complete. Just look at it as yet another example of dealing competently with ambiguous and incomplete data (which is, at root, all that intelligence is). You use your personal definition of natural language. I don't think that human's are intelligent because they use an ambiguous language. They also would be intelligent if their language would not suffer from ambiguities. One thought module, two translation modules -- except that all the translation modules really are is label appliers and grammar re-arrangers. The heavy lifting is all in the thought module. The problem is that you are claiming that language lies entirely in the translation modules while I'm arguing that a large percentage of it is in the thought module. The fact that the translation module has to go to the thought module for disambiguation and interpretation (and numerous other things) should make it quite clear that language is *not* simply translation. It is still just translation. Further, if you read Pinker's book, you will find that languages have a lot more in common than you would expect if language truly were independent of and separate from thought (as you are claiming). Language is built on top of the thinking/cognitive architecture (not beside it and not independent of it) and could not exist without it. That is why language is AGI-complete. Language also gives an excellent window into many of the features of that cognitive architecture and determining what is necessary for language also determine what is in that cognitive architecture. Another excellent window is how humans perform moral judgments (try reading Marc Hauser -- either his numerous scientific papers or the excellent Moral Minds). Or, yet another, is examining the structure of human biases. There are also visual thoughts. You can imagine objects moving. The principle is the same as with thoughts you perceive in your language: There is an internal representation of patterns which is completely hidden for your consciousness. The brain compresses and translates your visual thoughts and routes the results to its own visual input regions. As long as there is no real evidence against the model that thoughts are separated from the way I perceive thoughts (e.g. by language )I do not see any reason to change my opinion. - Matthias --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
AW: [agi] Words vs Concepts [ex Defining AGI]
I have given the example with the dog next to a tree. There is an ambiguity. It can be resolved because the pattern for dog has a stronger relation to the pattern for angry than it is the case for the pattern of tree. You don't have to manipulate any patterns and can do the translation. - Matthias Marc Walser wrote: How do you communicate something for which you have no established communications protocol? If you can answer that, you have solved the natural language problem. --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: [agi] Re: Defining AGI
Manipulating of patterns needs reading and writing operations. Data structures will be changed. Translation needs just reading operations to the patterns of the internal model. So translation is a pattern manipulation where the result isn't stored? I disagree that AGI must have some process for learning language. If we concentrate just on the domain of mathematics we could give AGI all the rules for a sufficient language to express its results and to understand our questions. The domain of mathematics is complete and unambiguous. A mathematics AI is not a GI in my book. It won't generalize to the real world until it handles incompleteness and ambiguity (which is my objection to your main analogy). (Note: I'm not saying that it might not be a good first step . . . . but I don't believe that it is on the shortest path to GI). New definitions makes communication more comfortable but they are not necessary. Wrong. Methane is not a new definition, it is a new label. New definitions that combine lots of raw data into much more manipulable knowledge are necessary exactly as much as a third-, fourth-, or fifth- generation language is necessary instead of machine language. I don't know the telephone game. The details are essential. It is not essential where the data comes from and where it ends. Just the process of translating internal data into a certain language and vice versa is important. Start with a circle of people. Tell the first person a reasonable length phrase, have them tell the next, and so on. The end result is fascinating and very similar to what happens when an incompetent pair of translators attempt to translate from one language to another and back again. It is clear that an AGI needs an interface for human beings. But the question in this discussion is whether the language interface is a key point in AGI or not. In my opinion it is no key point. It is just a communication protocol. The real intelligence has nothing to do with language understanding. Therefore we should use a simple formal hard coded language for first AGI. The communication protocol needs to be extensible to handle output after learning or transition into a new domain. How do you ground new concepts? More importantly, it needs to be extensible to support teaching the AGI. As I keep saying, how are you going to make your communication protocol extensible? Real GENERAL intelligence has EVERYTHING to do with extensibility. I don't see any problems with my model and I do not see any flaws which I don't have answered. I haven't seen any point where my analogy comes short. I keep pointing out that your model separating communication and database updating depends upon a fully specified model and does not tolerate ambiguity (i.e. it lacks extensibility and doesn't handle ambiguity). You continue not to answer these points. Unless you can handle valid objections by showing why they aren't valid, your model is disproven by counter-example. - Original Message - From: Dr. Matthias Heger To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 4:53 PM Subject: AW: AW: [agi] Re: Defining AGI Mark Waser wrote: How is translating patterns into language different from manipulating patterns? It seems to me that they are *exactly* the same thing. How do you believe that they differ? Manipulating of patterns needs reading and writing operations. Data structures will be changed. Translation needs just reading operations to the patterns of the internal model. Do you really believe that if A is easier than B then that makes A easy? How about if A is leaping a tall building in a single bound and B is jumping to the moon? The word *easy* is not exactly definable. Do you believe that language is fully specified? That we can program English into an AGI by hand? No. That's the reason why I would not use human language for the first AGI. Yes, I imagine that an AGI must have some process for learning language because language is necessary for learning knowledge and knowledge is necessary for intelligence. What part of that do you disagree with? Please be specific. I disagree that AGI must have some process for learning language. If we concentrate just on the domain of mathematics we could give AGI all the rules for a sufficient language to express its results and to understand our questions. And this is where we are not communicating. Since language is not fully specified, then the participants in many conversations are *constantly* creating and learning language as a part of the process of communication. This is where Gödel's incompleteness comes in. To be a General Intelligence, you must be able to extend beyond what is currently known and specified into new domains. Any time that we
Re: AW: AW: [agi] Re: Defining AGI
Funny, Ben. So . . . . could you clearly state why science can't be done by anyone willing to simply follow the recipe? Is it really anything other than the fact that they are stopped by their unconscious beliefs and biases? If so, what? Instead of a snide comment, defend your opinion with facts, explanations, and examples of what it really is. I can give you all sorts of examples where someone is capable of doing something by the numbers until they are told that they can't. What do you believe is so difficult about science other than overcoming the sub/unconscious? Your statement is obviously spoken by someone who has lectured as opposed to taught. - Original Message - From: Ben Goertzel To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 5:26 PM Subject: Re: AW: AW: [agi] Re: Defining AGI *Any* human who can understand language beyond a certain point (say, that of a slightly sub-average human IQ) can easily be taught to be a good scientist if they are willing to play along. Science is a rote process that can be learned and executed by anyone -- as long as their beliefs and biases don't get in the way. This is obviously spoken by someone who has never been a professional teacher ;-p ben g -- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] Words vs Concepts [ex Defining AGI]
I have given the example with the dog next to a tree. There is an ambiguity. It can be resolved because the pattern for dog has a stronger relation to the pattern for angry than it is the case for the pattern of tree. So, are the relationships between the various patterns in your translation module or in your cognitive module? I would argue that they are in your cognitive module. If you disagree, then I'll just agree to disagree because in order for them to be in your translation module then you'll have to be constantly updating your translation module which then contradicts what you said previously about the translation module being static. - Original Message - From: Dr. Matthias Heger [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 5:38 PM Subject: AW: [agi] Words vs Concepts [ex Defining AGI] I have given the example with the dog next to a tree. There is an ambiguity. It can be resolved because the pattern for dog has a stronger relation to the pattern for angry than it is the case for the pattern of tree. You don't have to manipulate any patterns and can do the translation. - Matthias Marc Walser wrote: How do you communicate something for which you have no established communications protocol? If you can answer that, you have solved the natural language problem. --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: AW: [agi] Re: Defining AGI
Mark, It is not the case that I have merely lectured rather than taught. I've lectured (math, CS, psychology and futurology) at university, it's true ... but I've also done extensive one-on-one math tutoring with students at various levels ... and I've also taught small groups of kids aged 7-12, hands-on (math programming), and I've taught retirees various skills (mostly computer related). Why can't a stupid person do good science? Doing science in reality seems to require a whole bunch of implicit, hard-to-verbalize knowledge that stupid people just don't seem to be capable of learning. A stupid person can possibly be trained to be a good lab assistant, in some areas of science but not others (it depends on how flaky and how complex the lab technology involved is in that area). But, being a scientist involves a lot of judgment, a lot of heuristic, uncertain reasoning drawing on a wide variety of knowledge. Could a stupid person learn to be a good scientist given, say, a thousand years of training? Maybe. But I doubt it, because by the time they had moved on to learning the second half of what they need to know, they would have already forgotten the first half ;-p You work in software engineering -- do you think a stupid person could be trained to be a really good programmer? Again, I very much doubt it ... though they could be (and increasingly are ;-p) trained to do routine programming tasks. Inevitably, in either of these cases, the person will encounter some situation not directly covered in their training, and will need to intelligently analogize to their experience, and will fail at this because they are not very intelligent... -- Ben G On Sun, Oct 19, 2008 at 5:43 PM, Mark Waser [EMAIL PROTECTED] wrote: Funny, Ben. So . . . . could you clearly state why science can't be done by anyone willing to simply follow the recipe? Is it really anything other than the fact that they are stopped by their unconscious beliefs and biases? If so, what? Instead of a snide comment, defend your opinion with facts, explanations, and examples of what it really is. I can give you all sorts of examples where someone is capable of doing something by the numbers until they are told that they can't. What do you believe is so difficult about science other than overcoming the sub/unconscious? Your statement is obviously spoken by someone who has lectured as opposed to taught. - Original Message - *From:* Ben Goertzel [EMAIL PROTECTED] *To:* agi@v2.listbox.com *Sent:* Sunday, October 19, 2008 5:26 PM *Subject:* Re: AW: AW: [agi] Re: Defining AGI *Any* human who can understand language beyond a certain point (say, that of a slightly sub-average human IQ) can easily be taught to be a good scientist if they are willing to play along. Science is a rote process that can be learned and executed by anyone -- as long as their beliefs and biases don't get in the way. This is obviously spoken by someone who has never been a professional teacher ;-p ben g -- *agi* | Archives https://www.listbox.com/member/archive/303/=now https://www.listbox.com/member/archive/rss/303/ | Modifyhttps://www.listbox.com/member/?;Your Subscription http://www.listbox.com -- *agi* | Archives https://www.listbox.com/member/archive/303/=now https://www.listbox.com/member/archive/rss/303/ | Modifyhttps://www.listbox.com/member/?;Your Subscription http://www.listbox.com -- Ben Goertzel, PhD CEO, Novamente LLC and Biomind LLC Director of Research, SIAI [EMAIL PROTECTED] Nothing will ever be attempted if all possible objections must be first overcome - Dr Samuel Johnson --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: AW: [agi] Re: Defining AGI
Interesting how you always only address half my points . . . I keep hammering extensibility and you focus on ambiguity which is merely the result of extensibility. You refuse to address extensibility. Maybe because it really is the secret sauce of intelligence and the one thing that you can't handle? And after a long explanation, I get comments like It is still just translation with no further explanation and visual thought nonsense worthy of Mike Tintner. So, I give up. I can't/won't debate someone who won't follow scientific methods of inquiry. - Original Message - From: Dr. Matthias Heger [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 5:21 PM Subject: AW: AW: AW: [agi] Re: Defining AGI Marc Walser wrote: *Any* human who can understand language beyond a certain point (say, that of a slightly sub-average human IQ) can easily be taught to be a good scientist if they are willing to play along. Science is a rote process that can be learned and executed by anyone -- as long as their beliefs and biases don't get in the way. This is just an opinion and I strongly disagree with your opinion. Obviously you overestimate language understanding a lot. This is a bit of disingenuous side-track that I feel that I must address. When people say natural language, the important features are extensibility and ambiguity. If you can handle one extensible and ambiguous language, you should have the capabilities to handle all of them. It's yet another definition of GI-complete. Just look at it as yet another example of dealing competently with ambiguous and incomplete data (which is, at root, all that intelligence is). You use your personal definition of natural language. I don't think that human's are intelligent because they use an ambiguous language. They also would be intelligent if their language would not suffer from ambiguities. One thought module, two translation modules -- except that all the translation modules really are is label appliers and grammar re-arrangers. The heavy lifting is all in the thought module. The problem is that you are claiming that language lies entirely in the translation modules while I'm arguing that a large percentage of it is in the thought module. The fact that the translation module has to go to the thought module for disambiguation and interpretation (and numerous other things) should make it quite clear that language is *not* simply translation. It is still just translation. Further, if you read Pinker's book, you will find that languages have a lot more in common than you would expect if language truly were independent of and separate from thought (as you are claiming). Language is built on top of the thinking/cognitive architecture (not beside it and not independent of it) and could not exist without it. That is why language is AGI-complete. Language also gives an excellent window into many of the features of that cognitive architecture and determining what is necessary for language also determine what is in that cognitive architecture. Another excellent window is how humans perform moral judgments (try reading Marc Hauser -- either his numerous scientific papers or the excellent Moral Minds). Or, yet another, is examining the structure of human biases. There are also visual thoughts. You can imagine objects moving. The principle is the same as with thoughts you perceive in your language: There is an internal representation of patterns which is completely hidden for your consciousness. The brain compresses and translates your visual thoughts and routes the results to its own visual input regions. As long as there is no real evidence against the model that thoughts are separated from the way I perceive thoughts (e.g. by language )I do not see any reason to change my opinion. - Matthias --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: AW: [agi] Re: Defining AGI
Actually, I should have drawn a distinction . . . . there is a major difference between performing discovery as a scientist and evaluating data as a scientist. I was referring to the latter (which is similar to understanding Einstein) as opposed to the former (which is being Einstein). You clearly are referring to the creative act of discovery (Programming is also a discovery operation). So let me rephrase my statement -- Can a stupid person do good scientific evaluation if taught the rules and willing to abide by them? Why or why not? - Original Message - From: Ben Goertzel To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 5:52 PM Subject: Re: AW: AW: [agi] Re: Defining AGI Mark, It is not the case that I have merely lectured rather than taught. I've lectured (math, CS, psychology and futurology) at university, it's true ... but I've also done extensive one-on-one math tutoring with students at various levels ... and I've also taught small groups of kids aged 7-12, hands-on (math programming), and I've taught retirees various skills (mostly computer related). Why can't a stupid person do good science? Doing science in reality seems to require a whole bunch of implicit, hard-to-verbalize knowledge that stupid people just don't seem to be capable of learning. A stupid person can possibly be trained to be a good lab assistant, in some areas of science but not others (it depends on how flaky and how complex the lab technology involved is in that area). But, being a scientist involves a lot of judgment, a lot of heuristic, uncertain reasoning drawing on a wide variety of knowledge. Could a stupid person learn to be a good scientist given, say, a thousand years of training? Maybe. But I doubt it, because by the time they had moved on to learning the second half of what they need to know, they would have already forgotten the first half ;-p You work in software engineering -- do you think a stupid person could be trained to be a really good programmer? Again, I very much doubt it ... though they could be (and increasingly are ;-p) trained to do routine programming tasks. Inevitably, in either of these cases, the person will encounter some situation not directly covered in their training, and will need to intelligently analogize to their experience, and will fail at this because they are not very intelligent... -- Ben G On Sun, Oct 19, 2008 at 5:43 PM, Mark Waser [EMAIL PROTECTED] wrote: Funny, Ben. So . . . . could you clearly state why science can't be done by anyone willing to simply follow the recipe? Is it really anything other than the fact that they are stopped by their unconscious beliefs and biases? If so, what? Instead of a snide comment, defend your opinion with facts, explanations, and examples of what it really is. I can give you all sorts of examples where someone is capable of doing something by the numbers until they are told that they can't. What do you believe is so difficult about science other than overcoming the sub/unconscious? Your statement is obviously spoken by someone who has lectured as opposed to taught. - Original Message - From: Ben Goertzel To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 5:26 PM Subject: Re: AW: AW: [agi] Re: Defining AGI *Any* human who can understand language beyond a certain point (say, that of a slightly sub-average human IQ) can easily be taught to be a good scientist if they are willing to play along. Science is a rote process that can be learned and executed by anyone -- as long as their beliefs and biases don't get in the way. This is obviously spoken by someone who has never been a professional teacher ;-p ben g -- agi | Archives | Modify Your Subscription agi | Archives | Modify Your Subscription -- Ben Goertzel, PhD CEO, Novamente LLC and Biomind LLC Director of Research, SIAI [EMAIL PROTECTED] Nothing will ever be attempted if all possible objections must be first overcome - Dr Samuel Johnson -- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: AW: [agi] Re: Defining AGI
Whether a stupid person can do good scientific evaluation if taught the rules is a badly-formed question, because no one knows what the rules are. They are learned via experience just as much as by explicit teaching Furthermore, as anyone who has submitted a lot of science papers to journals knows, even smart scientists can be horrendously bad at scientific evaluation. I've had some really good bioscience papers rejected from journals, by presumably intelligent referees, for extremely bad reasons (and these papers were eventually published in good journals). Evaluating research is not much easier than doing it. When is someone's supposed test of statistical validity really the right test? Too many biology referees just look for the magic number of p.05 rather than understanding what test actually underlies that number, because they don't know the math or don't know how to connect the math to the experiment in a contextually appropriate way. As another example: When should a data point be considered an outlier (meaning: probably due to equipment error or some other quirk) rather than a genuine part of the data? Tricky. I recall Feynman noting that he was held back in making a breakthrough discovery for some time, because of an outlier on someone else's published data table, which turned out to be spurious but had been accepted as valid by the community. In this case, Feyman's exceptional intelligence allowed him to carry out scientific evaluation more effectively than other, intelligent but less-so-than-him, had done... -- Ben G On Sun, Oct 19, 2008 at 6:00 PM, Mark Waser [EMAIL PROTECTED] wrote: Actually, I should have drawn a distinction . . . . there is a major difference between performing discovery as a scientist and evaluating data as a scientist. I was referring to the latter (which is similar to understanding Einstein) as opposed to the former (which is being Einstein). You clearly are referring to the creative act of discovery (Programming is also a discovery operation). So let me rephrase my statement -- Can a stupid person do good scientific evaluation if taught the rules and willing to abide by them? Why or why not? - Original Message - *From:* Ben Goertzel [EMAIL PROTECTED] *To:* agi@v2.listbox.com *Sent:* Sunday, October 19, 2008 5:52 PM *Subject:* Re: AW: AW: [agi] Re: Defining AGI Mark, It is not the case that I have merely lectured rather than taught. I've lectured (math, CS, psychology and futurology) at university, it's true ... but I've also done extensive one-on-one math tutoring with students at various levels ... and I've also taught small groups of kids aged 7-12, hands-on (math programming), and I've taught retirees various skills (mostly computer related). Why can't a stupid person do good science? Doing science in reality seems to require a whole bunch of implicit, hard-to-verbalize knowledge that stupid people just don't seem to be capable of learning. A stupid person can possibly be trained to be a good lab assistant, in some areas of science but not others (it depends on how flaky and how complex the lab technology involved is in that area). But, being a scientist involves a lot of judgment, a lot of heuristic, uncertain reasoning drawing on a wide variety of knowledge. Could a stupid person learn to be a good scientist given, say, a thousand years of training? Maybe. But I doubt it, because by the time they had moved on to learning the second half of what they need to know, they would have already forgotten the first half ;-p You work in software engineering -- do you think a stupid person could be trained to be a really good programmer? Again, I very much doubt it ... though they could be (and increasingly are ;-p) trained to do routine programming tasks. Inevitably, in either of these cases, the person will encounter some situation not directly covered in their training, and will need to intelligently analogize to their experience, and will fail at this because they are not very intelligent... -- Ben G On Sun, Oct 19, 2008 at 5:43 PM, Mark Waser [EMAIL PROTECTED] wrote: Funny, Ben. So . . . . could you clearly state why science can't be done by anyone willing to simply follow the recipe? Is it really anything other than the fact that they are stopped by their unconscious beliefs and biases? If so, what? Instead of a snide comment, defend your opinion with facts, explanations, and examples of what it really is. I can give you all sorts of examples where someone is capable of doing something by the numbers until they are told that they can't. What do you believe is so difficult about science other than overcoming the sub/unconscious? Your statement is obviously spoken by someone who has lectured as opposed to taught. - Original Message - *From:* Ben Goertzel [EMAIL PROTECTED] *To:* agi@v2.listbox.com *Sent:* Sunday, October 19, 2008 5:26 PM *Subject:*
Re: AW: AW: [agi] Re: Defining AGI
Whether a stupid person can do good scientific evaluation if taught the rules is a badly-formed question, because no one knows what the rules are. They are learned via experience just as much as by explicit teaching Wow! I'm sorry but that is a very scary, incorrect opinion. There's a really good book called The Game of Science by McCain and Segal that clearly explains all of the rules. I'll get you a copy. I understand that most scientists aren't trained properly -- but that is no reason to continue the problem by claiming that they can't be trained properly. You make my point with your explanation of your example of biology referees. And the Feynman example, if it is the story that I've heard before, was actually an example of good science in action because the outlier was eventually overruled AFTER ENOUGH GOOD DATA WAS COLLECTED to prove that the outlier was truly an outlier and not just a mere inconvenience to someone's theory. Feynman's exceptional intelligence allowed him to discover a possibility that might have been correct if the point was an outlier, but good scientific evaluation relies on data, data, and more data. Using that story as an example shows that you don't understand how to properly run a scientific evaluative process. - Original Message - From: Ben Goertzel To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 6:07 PM Subject: Re: AW: AW: [agi] Re: Defining AGI Whether a stupid person can do good scientific evaluation if taught the rules is a badly-formed question, because no one knows what the rules are. They are learned via experience just as much as by explicit teaching Furthermore, as anyone who has submitted a lot of science papers to journals knows, even smart scientists can be horrendously bad at scientific evaluation. I've had some really good bioscience papers rejected from journals, by presumably intelligent referees, for extremely bad reasons (and these papers were eventually published in good journals). Evaluating research is not much easier than doing it. When is someone's supposed test of statistical validity really the right test? Too many biology referees just look for the magic number of p.05 rather than understanding what test actually underlies that number, because they don't know the math or don't know how to connect the math to the experiment in a contextually appropriate way. As another example: When should a data point be considered an outlier (meaning: probably due to equipment error or some other quirk) rather than a genuine part of the data? Tricky. I recall Feynman noting that he was held back in making a breakthrough discovery for some time, because of an outlier on someone else's published data table, which turned out to be spurious but had been accepted as valid by the community. In this case, Feyman's exceptional intelligence allowed him to carry out scientific evaluation more effectively than other, intelligent but less-so-than-him, had done... -- Ben G On Sun, Oct 19, 2008 at 6:00 PM, Mark Waser [EMAIL PROTECTED] wrote: Actually, I should have drawn a distinction . . . . there is a major difference between performing discovery as a scientist and evaluating data as a scientist. I was referring to the latter (which is similar to understanding Einstein) as opposed to the former (which is being Einstein). You clearly are referring to the creative act of discovery (Programming is also a discovery operation). So let me rephrase my statement -- Can a stupid person do good scientific evaluation if taught the rules and willing to abide by them? Why or why not? - Original Message - From: Ben Goertzel To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 5:52 PM Subject: Re: AW: AW: [agi] Re: Defining AGI Mark, It is not the case that I have merely lectured rather than taught. I've lectured (math, CS, psychology and futurology) at university, it's true ... but I've also done extensive one-on-one math tutoring with students at various levels ... and I've also taught small groups of kids aged 7-12, hands-on (math programming), and I've taught retirees various skills (mostly computer related). Why can't a stupid person do good science? Doing science in reality seems to require a whole bunch of implicit, hard-to-verbalize knowledge that stupid people just don't seem to be capable of learning. A stupid person can possibly be trained to be a good lab assistant, in some areas of science but not others (it depends on how flaky and how complex the lab technology involved is in that area). But, being a scientist involves a lot of judgment, a lot of heuristic, uncertain reasoning drawing on a wide variety of knowledge. Could a stupid person learn to be a good scientist given, say, a thousand years of training? Maybe. But I doubt it, because by the time they had moved
Re: [agi] constructivist issues
Matt, Yes, that is completely true. I should have worded myself more clearly. Ben, Matt has sorted out the mistake you are referring to. What I meant was that AIXI is incapable of understanding the proof, not that it is incapable of producing it. Another way of describing it: AIXI could learn to accurately mimic the way humans talk about uncomputable entities, but it would never invent these things on its own. --Abram On Sun, Oct 19, 2008 at 4:32 PM, Matt Mahoney [EMAIL PROTECTED] wrote: --- On Sat, 10/18/08, Abram Demski [EMAIL PROTECTED] wrote: No, I do not claim that computer theorem-provers cannot prove Goedel's Theorem. It has been done. The objection applies specifically to AIXI-- AIXI cannot prove goedel's theorem. Yes it can. It just can't understand its own proof in the sense of Tarski's undefinability theorem. Construct a predictive AIXI environment as follows: the environment output symbol does not depend on anything the agent does. However, the agent receives a reward when its output symbol matches the next symbol input from the environment. Thus, the environment can be modeled as a string that the agent has the goal of compressing. Now encode in the environment a series of theorems followed by their proofs. Since proofs can be mechanically checked, and therefore found given enough time (if the proof exists), then the optimal strategy for the agent, according to AIXI is to guess that the environment receives as input a series of theorems and that the environment then proves them and outputs the proof. AIXI then replicates its guess, thus correctly predicting the proofs and maximizing its reward. To prove Goedel's theorem, we simply encode it into the environment after a series of other theorems and their proofs. -- Matt Mahoney, [EMAIL PROTECTED] --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: [agi] Re: Defining AGI
Hmm. After the recent discussion it seems this list has turned into the philosophical musings related to AGI list. Where is the AGI engineering list? - samantha --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: AW: [agi] Re: Defining AGI
Sorry Mark, but I'm not going to accept your opinion on this just because you express it with vehemence and confidence. I didn't argue much previously when you told me I didn't understand engineering ... because, although I've worked with a lot of engineers, I haven't been one. But, I grew up around scientists, I've trained scientists, and I am currently (among other things) working as a scientist. It is really not true that there is a set of simple rules adequate to tell people how to evaluate scientific results effectively. As often occurs, there may be rules that tell you how to handle 80% of cases (or whatever), but then the remainder of the cases are harder and require actual judgment. This is, by the way, the case with essentially every complex human process that people have sought to cover via expert rules. The rules cover many cases ... but as one seeks to extend them to cover all relevant cases, one winds up adding more and more and more specialized rules... It is possible I inaccurately remembered an anecdote from Feynman's book, but that's irrelevant to my point. *** Using that story as an example shows that you don't understand how to properly run a scientific evaluative process. *** Wow, that is quite an insult. So you're calling me an incompetent in my profession now. I don't have particularly thin skin, but I have to say that I'm getting really tired of being attacked and insulted on this email list. -- Ben G On Sun, Oct 19, 2008 at 6:18 PM, Mark Waser [EMAIL PROTECTED] wrote: Whether a stupid person can do good scientific evaluation if taught the rules is a badly-formed question, because no one knows what the rules are. They are learned via experience just as much as by explicit teaching Wow! I'm sorry but that is a very scary, incorrect opinion. There's a really good book called The Game of Science by McCain and Segal that clearly explains all of the rules. I'll get you a copy. I understand that most scientists aren't trained properly -- but that is no reason to continue the problem by claiming that they can't be trained properly. You make my point with your explanation of your example of biology referees. And the Feynman example, if it is the story that I've heard before, was actually an example of good science in action because the outlier was eventually overruled AFTER ENOUGH GOOD DATA WAS COLLECTED to prove that the outlier was truly an outlier and not just a mere inconvenience to someone's theory. Feynman's exceptional intelligence allowed him to discover a possibility that might have been correct if the point was an outlier, but good scientific evaluation relies on data, data, and more data. Using that story as an example shows that you don't understand how to properly run a scientific evaluative process. - Original Message - *From:* Ben Goertzel [EMAIL PROTECTED] *To:* agi@v2.listbox.com *Sent:* Sunday, October 19, 2008 6:07 PM *Subject:* Re: AW: AW: [agi] Re: Defining AGI Whether a stupid person can do good scientific evaluation if taught the rules is a badly-formed question, because no one knows what the rules are. They are learned via experience just as much as by explicit teaching Furthermore, as anyone who has submitted a lot of science papers to journals knows, even smart scientists can be horrendously bad at scientific evaluation. I've had some really good bioscience papers rejected from journals, by presumably intelligent referees, for extremely bad reasons (and these papers were eventually published in good journals). Evaluating research is not much easier than doing it. When is someone's supposed test of statistical validity really the right test? Too many biology referees just look for the magic number of p.05 rather than understanding what test actually underlies that number, because they don't know the math or don't know how to connect the math to the experiment in a contextually appropriate way. As another example: When should a data point be considered an outlier (meaning: probably due to equipment error or some other quirk) rather than a genuine part of the data? Tricky. I recall Feynman noting that he was held back in making a breakthrough discovery for some time, because of an outlier on someone else's published data table, which turned out to be spurious but had been accepted as valid by the community. In this case, Feyman's exceptional intelligence allowed him to carry out scientific evaluation more effectively than other, intelligent but less-so-than-him, had done... -- Ben G On Sun, Oct 19, 2008 at 6:00 PM, Mark Waser [EMAIL PROTECTED] wrote: Actually, I should have drawn a distinction . . . . there is a major difference between performing discovery as a scientist and evaluating data as a scientist. I was referring to the latter (which is similar to understanding Einstein) as opposed to the former (which is being Einstein). You clearly are referring to
Re: AW: [agi] Re: Defining AGI
I've been on some message boards where people only ever came back with a formula or a correction. I didn't contribute a great deal but it is a sight for sore eyes. We could have an agi-tech and an agi-philo list and maybe they'd merit further recombination (more lists) after that. --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
AW: [agi] Words vs Concepts [ex Defining AGI]
Absolutely. We are not aware of most of our assumptions that are based in our common heritage, culture, and embodiment. But an external observer could easily notice them and tease out an awful lot of information about us by doing so. You do not understand what I mean. There will be lot of implementation details (e.g. temporary variables ) within the patterns which will never be send by linguistic messages. I disagree with a complete distinction between D and L. L is a very small fraction of D translated for transmission. However, instead of arguing that there must be a strict separation between language model and D, I would argue that the more similar the two could be (i.e. the less translation from D to L) the better. Analyzing L in that case could tell you more about D than you might think (which is what Pinker and Hauser argue). It's like looking at data to determine an underlying cause for a phenomenon. Even noticing what does and does not vary (and what covaries) tells you a lot about the underlying cause (D). This is just an assumption of you. No facts. My opinion remains: D and L are separated. How do you go from a formal language to a competent description of a messy, ambiguous, data-deficient world? *That* is the natural language question. Any algorithm in your computer is written in a formal well defined language. If you agree that AGI is possible with current programming languages then you have to agree that the ambiguous, data-deficient world can be managed by formal languages. What happens if I say that language extensibility is exactly analogous to learning which is exactly analogous to internal model improvement? What happens? I disagree. So translation is a pattern manipulation where the result isn't stored? The result isn't stored in D The domain of mathematics is complete and unambiguous. A mathematics AI is not a GI in my book. It won't generalize to the real world until it handles incompleteness and ambiguity (which is my objection to your main analogy). If you say mathematics is not GI then the following must be true for you: The universe cannot be modeled by mathematics. I disagree. The communication protocol needs to be extensible to handle output after learning or transition into a new domain. How do you ground new concepts? More importantly, it needs to be extensible to support teaching the AGI. As I keep saying, how are you going to make your communication protocol extensible? Real GENERAL intelligence has EVERYTHING to do with extensibility. For mathematics you just need a few axioms. There are an infinite number of expressions which can be written with a final set of symbols and a finite formal language. But extensibility is no crucial point in this discussion at all. You can have extensibility with a strict separation of D and L. For first AGI with mathematics I would hardcode an algorithm which manages an open list of axioms and definitions as a language interface. I keep pointing out that your model separating communication and database updating depends upon a fully specified model and does not tolerate ambiguity (i.e. it lacks extensibility and doesn't handle ambiguity). You continue not to answer these points. Once again: The separation of communication and database updating does not contradict extensibility and ambiguity. Language data and domain data can be strictly separated. I can update the language database without communicating (e.g. just by changing the hard disk) or with communicating. The main point is that the model *D* needs not to be changed during communicating. Furthermore, language extension would be a nice feature but it is not necessary. The model D needs not to be fully specified at all. If the model L is formal and without ambiguities this does not imply at all that problems with ambiguities cannot be handled. --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: AW: [agi] Re: Defining AGI
Mark, I did not say that theory should trump data. When theory should trump data is a very complex question. I don't mind reading the book you suggested eventually but I have a long list of other stuff to read that seems to have higher priority. I don't believe there exists a complete, well-defined set of rules for evaluating scientific evidence in real-world cases, sorry. If you want to say there is a complete set of rules, but there is no complete set of rules for how to apply these rules -- well, I still doubt it, but it seems like a less absurd contention. But in that case, so what? In that case the rules don't actually tell you how to evaluate scientific evidence. In bioinformatics, it seems to me that evaluating complex datasets gets into tricky questions of applied statistics, on which expert biostatisticians don't always agree (and they write papers about their arguments). Clearly this is not a pursuit for dumbheads ;-p ... Perhaps you would classify this as a dispute about how to apply the rules and not about the rules themselves? I don't really understand the distinction you're drawing there... ben g On Sun, Oct 19, 2008 at 7:15 PM, Mark Waser [EMAIL PROTECTED] wrote: It is really not true that there is a set of simple rules adequate to tell people how to evaluate scientific results effectively. Get the book and then speak from a position of knowledge by telling me something that you believe it is missing. When I cite a specific example that you can go and verify or disprove, it is not an opinion but a valid data point (and your perception of my vehemence and/or confidence and your personal reaction to it are totally irrelevant). The fact that you can make a statement like this from a position of total ignorance when I cite a specific example is a clear example of not following basic scientific principles. You can be insulted all you like but that is not what a good scientist would do on a good day -- it is simply lazy and bad science. As often occurs, there may be rules that tell you how to handle 80% of cases (or whatever), but then the remainder of the cases are harder and require actual judgment. Is it that the rules don't have 100% coverage or is that it isn't always clear how to appropriately apply the rules and that is where the questions come in? There is a huge difference between the two cases -- and your statement no one knows what the rules are argues for the former not the latter. I'd be more than willing to accept the latter -- but the former is an embarrassment. Do you really mean to contend the former? It is possible I inaccurately remembered an anecdote from Feynman's book, but that's irrelevant to my point. No, you accurately remembered the anecdote. As I recall, Feynman was expressing frustration at the slowness of the process -- particularly because no one would consider his hypothesis enough to perform the experiments necessary to determine whether the point was an outlier or not. Not performing the experiment was an unfortunate choice of trade-offs (since I'm sure that they were doing something else that they deemed more likely to produce worthwhile results) but accepting his theory without first proving that the outlier was indeed an outlier (regardless of his intelligence) would have been far worse and directly contrary to the scientific method. Using that story as an example shows that you don't understand how to properly run a scientific evaluative process. Wow, that is quite an insult. So you're calling me an incompetent in my profession now. It depends. Are you going to continue promoting something as inexcusable as saying that theory should trump data (because of the source of the theory)? I was quite clear that I was criticizing a very specific action. Are you going to continue to defend that improper action? And why don't we keep this on the level of scientific debate rather than arguing insults and vehemence and confidence? That's not particularly good science either. - Original Message - *From:* Ben Goertzel [EMAIL PROTECTED] *To:* agi@v2.listbox.com *Sent:* Sunday, October 19, 2008 6:31 PM *Subject:* Re: AW: AW: [agi] Re: Defining AGI Sorry Mark, but I'm not going to accept your opinion on this just because you express it with vehemence and confidence. I didn't argue much previously when you told me I didn't understand engineering ... because, although I've worked with a lot of engineers, I haven't been one. But, I grew up around scientists, I've trained scientists, and I am currently (among other things) working as a scientist. It is really not true that there is a set of simple rules adequate to tell people how to evaluate scientific results effectively. As often occurs, there may be rules that tell you how to handle 80% of cases (or whatever), but then the remainder of the cases are harder and require actual judgment. This is, by the way, the case
Re: AW: AW: [agi] Re: Defining AGI
And why don't we keep this on the level of scientific debate rather than arguing insults and vehemence and confidence? That's not particularly good science either. Right ... being unnecessarily nasty is not either good or bad science, it's just irritating for others to deal with ben g --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] Words vs Concepts [ex Defining AGI]
I disagree with a complete distinction between D and L. L is a very small fraction of D translated for transmission. However, instead of arguing that there must be a strict separation between language model and D, I would argue that the more similar the two could be (i.e. the less translation from D to L) the better. Analyzing L in that case could tell you more about D than you might think (which is what Pinker and Hauser argue). It's like looking at data to determine an underlying cause for a phenomenon. Even noticing what does and does not vary (and what covaries) tells you a lot about the underlying cause (D). This is just an assumption of you. No facts. My opinion remains: D and L are separated. Geez. What is it with this list? Read Pinker. Tons of facts. Take them into account and then form an opinion. Any algorithm in your computer is written in a formal well defined language. If you agree that AGI is possible with current programming languages then you have to agree that the ambiguous, data-deficient world can be managed by formal languages. Once we figure out how to program the process of automatically extending formal languages -- yes, absolutely. That's the path to AGI. If you say mathematics is not GI then the following must be true for you: The universe cannot be modeled by mathematics. I disagree. with Gödel? That's impressive. Furthermore, language extension would be a nice feature but it is not necessary. Cool. And this is where we agree to disagree (and it does seem to be at the root of all the other arguments). If I believed this, I would agree with most of your other stuff. I just don't see how you're going to stretch any non-extensible language to *effectively* cover an infinite universe. Gödel argues against it. - Original Message - From: Dr. Matthias Heger [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, October 19, 2008 6:53 PM Subject: AW: [agi] Words vs Concepts [ex Defining AGI] Absolutely. We are not aware of most of our assumptions that are based in our common heritage, culture, and embodiment. But an external observer could easily notice them and tease out an awful lot of information about us by doing so. You do not understand what I mean. There will be lot of implementation details (e.g. temporary variables ) within the patterns which will never be send by linguistic messages. I disagree with a complete distinction between D and L. L is a very small fraction of D translated for transmission. However, instead of arguing that there must be a strict separation between language model and D, I would argue that the more similar the two could be (i.e. the less translation from D to L) the better. Analyzing L in that case could tell you more about D than you might think (which is what Pinker and Hauser argue). It's like looking at data to determine an underlying cause for a phenomenon. Even noticing what does and does not vary (and what covaries) tells you a lot about the underlying cause (D). This is just an assumption of you. No facts. My opinion remains: D and L are separated. How do you go from a formal language to a competent description of a messy, ambiguous, data-deficient world? *That* is the natural language question. Any algorithm in your computer is written in a formal well defined language. If you agree that AGI is possible with current programming languages then you have to agree that the ambiguous, data-deficient world can be managed by formal languages. What happens if I say that language extensibility is exactly analogous to learning which is exactly analogous to internal model improvement? What happens? I disagree. So translation is a pattern manipulation where the result isn't stored? The result isn't stored in D The domain of mathematics is complete and unambiguous. A mathematics AI is not a GI in my book. It won't generalize to the real world until it handles incompleteness and ambiguity (which is my objection to your main analogy). If you say mathematics is not GI then the following must be true for you: The universe cannot be modeled by mathematics. I disagree. The communication protocol needs to be extensible to handle output after learning or transition into a new domain. How do you ground new concepts? More importantly, it needs to be extensible to support teaching the AGI. As I keep saying, how are you going to make your communication protocol extensible? Real GENERAL intelligence has EVERYTHING to do with extensibility. For mathematics you just need a few axioms. There are an infinite number of expressions which can be written with a final set of symbols and a finite formal language. But extensibility is no crucial point in this discussion at all. You can have extensibility with a strict separation of D and L. For first AGI with mathematics I would hardcode an algorithm which manages an open list of axioms and
Re: AW: [agi] Re: Defining AGI
I've been thinking. agi-phil might suffice. Although it isn't as explicit. On Sun, Oct 19, 2008 at 6:52 PM, Eric Burton [EMAIL PROTECTED] wrote: I've been on some message boards where people only ever came back with a formula or a correction. I didn't contribute a great deal but it is a sight for sore eyes. We could have an agi-tech and an agi-philo list and maybe they'd merit further recombination (more lists) after that. --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: AW: [agi] Re: Defining AGI
No, surely this is mostly outside the purview of the AGI list. I'm reading some of this material and not getting a lot out of it. There are channels on freenode for this stuff. But we have got to agree on something if we are going to do anything. Can animals do science? They can not. --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: AW: AW: [agi] Re: Defining AGI
OK, well, I'm not going to formally kill this irrelevant-to-AGI thread as moderator, but I'm going to abandon it as participant... Time to get some work done tonight, enough time spent on email ;-p ben g On Sun, Oct 19, 2008 at 7:52 PM, Eric Burton [EMAIL PROTECTED] wrote: No, surely this is mostly outside the purview of the AGI list. I'm reading some of this material and not getting a lot out of it. There are channels on freenode for this stuff. But we have got to agree on something if we are going to do anything. Can animals do science? They can not. --- 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 -- Ben Goertzel, PhD CEO, Novamente LLC and Biomind LLC Director of Research, SIAI [EMAIL PROTECTED] Nothing will ever be attempted if all possible objections must be first overcome - Dr Samuel Johnson --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] constructivist issues
Ben, How so? Also, do you think it is nonsensical to put some probability on noncomputable models of the world? --Abram On Sun, Oct 19, 2008 at 6:33 PM, Ben Goertzel [EMAIL PROTECTED] wrote: But: it seems to me that, in the same sense that AIXI is incapable of understanding proofs about uncomputable numbers, **so are we humans** ... On Sun, Oct 19, 2008 at 6:30 PM, Abram Demski [EMAIL PROTECTED] wrote: Matt, Yes, that is completely true. I should have worded myself more clearly. Ben, Matt has sorted out the mistake you are referring to. What I meant was that AIXI is incapable of understanding the proof, not that it is incapable of producing it. Another way of describing it: AIXI could learn to accurately mimic the way humans talk about uncomputable entities, but it would never invent these things on its own. --Abram On Sun, Oct 19, 2008 at 4:32 PM, Matt Mahoney [EMAIL PROTECTED] wrote: --- On Sat, 10/18/08, Abram Demski [EMAIL PROTECTED] wrote: No, I do not claim that computer theorem-provers cannot prove Goedel's Theorem. It has been done. The objection applies specifically to AIXI-- AIXI cannot prove goedel's theorem. Yes it can. It just can't understand its own proof in the sense of Tarski's undefinability theorem. Construct a predictive AIXI environment as follows: the environment output symbol does not depend on anything the agent does. However, the agent receives a reward when its output symbol matches the next symbol input from the environment. Thus, the environment can be modeled as a string that the agent has the goal of compressing. Now encode in the environment a series of theorems followed by their proofs. Since proofs can be mechanically checked, and therefore found given enough time (if the proof exists), then the optimal strategy for the agent, according to AIXI is to guess that the environment receives as input a series of theorems and that the environment then proves them and outputs the proof. AIXI then replicates its guess, thus correctly predicting the proofs and maximizing its reward. To prove Goedel's theorem, we simply encode it into the environment after a series of other theorems and their proofs. -- Matt Mahoney, [EMAIL PROTECTED] --- 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/?; Powered by Listbox: http://www.listbox.com -- Ben Goertzel, PhD CEO, Novamente LLC and Biomind LLC Director of Research, SIAI [EMAIL PROTECTED] Nothing will ever be attempted if all possible objections must be first overcome - Dr Samuel Johnson 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com
Re: [agi] constructivist issues
Ben, Just to clarify my opinion: I think an actual implementation of the novamente/OCP design is likely to overcome this difficulty. However, to the extent that it approximates AIXI, I think there will be problems of these sorts. The main reason I think OCP/novamente would *not* approximate AIXI is that these systems are capable of a greater degree of self-reference, as well as a very different sort of adaptation. Self-reference gives the system a very direct reason to think *about* processes (resulting in halting, convergence, and other uncomputable properties). Self-adaptation could allow the system to adopt new sorts of reasoning (such as uncomputable models) simply because they seem to work. (This is different from AIXI being trained to prove theorems about uncomputable things, because the system starts actually making use of the theorems internally.) If I could formalize that intuition, I would be happy. --Abram On Sun, Oct 19, 2008 at 9:33 PM, Abram Demski [EMAIL PROTECTED] wrote: Ben, How so? Also, do you think it is nonsensical to put some probability on noncomputable models of the world? --Abram --- 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=117534816-b15a34 Powered by Listbox: http://www.listbox.com