Re: [agi] The Advantages of a Conscious Mind
On 5/6/07, Mike Tintner [EMAIL PROTECTED] wrote: YKY: Consciousness is not central to AGI . The human mind consists of a two-tier structure. On top, you have this conscious, executive mind that takes most of the decisions about which way the system will go - basically does the steering. On bottom, you have the unconscious, subordinate mind that does nearly all the information processing, both briefing and executing the executive mind's decisions, putting the words in its mouth and forming the thoughts in its head, while continually pressuring the executive mind with conflicting emotions, and at the same time monitoring and controlling the immensely complex operations of the body. That sounds reasonable. You're talking about the executive / planner module. My focus is on the truth maintenance module, which operates somewhat passively, and would require high-level directives from the planner, including value-based bias. The executive should be able to control all other modules. I tried not to use the term emotion in AGI, but I guess most people like it as a metaphor. YKY - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] A New Approach to AGI: What to do and what not to do (includes my revised algorithm)
to find a word in a big list you should really use a dictionary / hash table instead of binary search... ;-) (ok i know that wasnt the point you were trying to make :) Jean-Paul PS: [META] - people pls to cut off long message includes - some of us don't enjoy always on high bandwidth :( a [EMAIL PROTECTED] 05/06/07 2:36 AM For example, in computational linguistics, the algorithm can use a binary search to find records relating to a word, instead of scanning the whole database. What I mean is that the database can use indexes with a binary search algorithm to locate the word faster. This means that it avoids scanning each and every record of the database to find the pixel representation of the letters of the word (the bitmap image of the word). - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Mike, The extent to which there is a rigid distinction between these two tiers in the human brain/mind is not entirely clear. The human brain seems to have some distinct memory subsystems associated with various sorts of short term memory or working memory, but the notion of executive processing overall is IMO best thought of as a fuzzy set. Yes, there are some parts of the brain clearly shown (by fMRI and PET) to be involved with overall coordination, but the knowledge/memories associated by these brain regions is not necessarily the totality of what can occur in subjective conscious awareness. I think that the working memory and the autonomic nervous system are best viewed as two extremes, with a continuum of conscious intensity levels existing between them. For relatively recent thinking on the underpinnings of consciousnes in the human brain, check out the edited volume -- Neural Correlates of Consciousness, by Thomas Metzinger His single-author book -- Being No One is also very good, though I disagree with his take on AI at the end of the book. (he argues it would be unethical to create AGI's because it would be unethical to experiment on their half-formed, probably buggy conscious minds.) In Novamente we do have an AttentionalFocus concept which is much like what you call the conscious tier. We have chosen the term attentional focus to avoid getting into arguments related to the nature of consciousness and the first person versus third person perspectives on mind. Each item in the attentional focus is associated with a distributed network of other items that are not necessarily in the attentional focus, which ties in with the fuzziness of the executive function as mentioned above. -- Ben G On 5/6/07, YKY (Yan King Yin) [EMAIL PROTECTED] wrote: On 5/6/07, Mike Tintner [EMAIL PROTECTED] wrote: YKY: Consciousness is not central to AGI . The human mind consists of a two-tier structure. On top, you have this conscious, executive mind that takes most of the decisions about which way the system will go - basically does the steering. On bottom, you have the unconscious, subordinate mind that does nearly all the information processing, both briefing and executing the executive mind's decisions, putting the words in its mouth and forming the thoughts in its head, while continually pressuring the executive mind with conflicting emotions, and at the same time monitoring and controlling the immensely complex operations of the body. That sounds reasonable. You're talking about the executive / planner module. My focus is on the truth maintenance module, which operates somewhat passively, and would require high-level directives from the planner, including value-based bias. The executive should be able to control all other modules. I tried not to use the term emotion in AGI, but I guess most people like it as a metaphor. YKY -- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Well, there obviously IS a conscious, executive mind, separate from the unconscious mind, whatever the enormous difficulties cognitive sicentists had in first admitting its existence and now in identifying its correlates! And you still seem to be sharing some of those old difficulties in talking about it. Science generally still has some of those difficulties too. They shouldn't be there. Social organizations have chief executives and appear more or less incapable of functioning without them. The individual organization that is a human being appears to need an executive mind for much the same reasons - though those reasons need defining. Note that Fodor acknowledges the embarrassing truth that sicence can currently offer no explanation of why the conscious mind exists - rational, deterministic computers and machines clearly do not have or need one, functioning perfectly as entirely unconscious affairs. One immediate reason, applicable to AGI - although it will take the next Cognitive Revolution to recognize this - is that the two minds, almost certainly, think very differently. The unconscious mind thinks more or less algorithmically, (at least most of the time), rapidly in set ways - like a rational computer - it has to. Its function is to get things done. The conscious mind thinks literally, freely. How long it will spend on any given decision, and what course of thought it will pursue in reaching that decision are definitely NOT set, but free. (How does Pei's NARS fit in here?) Should I buy the marshmallow or the creme caramel ice cream? Hmm that's a tough one. I want to get this right... And I could and will resolve that decision in a few more seconds OR at other times, I could still be here thinking about it several minutes later OR at other times I could wander off in mid-thought to another subject entirely. No computer currently thinks like this - thinks freely and crazily as opposed to rationally and deterministically. Anyone who produces one - that has a similar practicality to the animal/human executive mind - will literally usher in the next Cognitive Revolution. You guys are clearly moving that way - but still appear to have a somewhat confused philosophical understanding of why all this is really necessary. (One interesting, but tangential issue is that the unconscious mind does appear to have a certain freedom too - it's hard to see dreams, for example, as deterministic affairs, Well, your dreams maybe, but not mine, you understand...). - Original Message - From: Benjamin Goertzel To: agi@v2.listbox.com Sent: Sunday, May 06, 2007 10:37 AM Subject: Re: [agi] The Advantages of a Conscious Mind Mike, The extent to which there is a rigid distinction between these two tiers in the human brain/mind is not entirely clear. The human brain seems to have some distinct memory subsystems associated with various sorts of short term memory or working memory, but the notion of executive processing overall is IMO best thought of as a fuzzy set. Yes, there are some parts of the brain clearly shown (by fMRI and PET) to be involved with overall coordination, but the knowledge/memories associated by these brain regions is not necessarily the totality of what can occur in subjective conscious awareness. I think that the working memory and the autonomic nervous system are best viewed as two extremes, with a continuum of conscious intensity levels existing between them. For relatively recent thinking on the underpinnings of consciousnes in the human brain, check out the edited volume -- Neural Correlates of Consciousness, by Thomas Metzinger His single-author book -- Being No One is also very good, though I disagree with his take on AI at the end of the book. (he argues it would be unethical to create AGI's because it would be unethical to experiment on their half-formed, probably buggy conscious minds.) In Novamente we do have an AttentionalFocus concept which is much like what you call the conscious tier. We have chosen the term attentional focus to avoid getting into arguments related to the nature of consciousness and the first person versus third person perspectives on mind. Each item in the attentional focus is associated with a distributed network of other items that are not necessarily in the attentional focus, which ties in with the fuzziness of the executive function as mentioned above. -- Ben G On 5/6/07, YKY (Yan King Yin) [EMAIL PROTECTED] wrote: On 5/6/07, Mike Tintner [EMAIL PROTECTED] wrote: YKY: Consciousness is not central to AGI . The human mind consists of a two-tier structure. On top, you have this conscious, executive mind that takes most of the decisions about which way the system will go - basically does the steering. On bottom, you have the unconscious, subordinate mind that does nearly all the information processing, both briefing and
Re: [agi] The Advantages of a Conscious Mind
Mike, The conscious mind thinks literally, freely. How long it will spend on any given decision, and what course of thought it will pursue in reaching that decision are definitely NOT set, but free. Ah, well, I'm glad to see the age-old problem of free will versus determinism is solved now! Mike has spoken!! ;-) Seriously ... have you read Libet's work on free will and the brain? Have you read Dennett's book Freedom Evolves? How about The Illusion of Conscious Will? The illusion of free will is a pretty subtle issue. I have made my own hypothesis regarding the sort of mechanism that underlies it in the human mind/brain, which is described in my 2006 book the Hidden Pattern and in preliminary form here: http://www.goertzel.org/dynapsyc/2004/FreeWill.htm You guys are clearly moving that way - but still appear to have a somewhat confused philosophical understanding of why all this is really necessary. Mike ... really ... has it ever occurred to you that you might NOT have a deeper understanding of these issues than people who have read all the existing literature on the topics and thought about them for decades?? On some topics, naive intuition can be misleading. Especially topics that involve illusions we humans have **evolved** to hold intuitively, so as to make our lives simpler... Please note that the naive notion of freedom you advocate contradicts all known physics including quantum physics and (all currently seriously debated variants of) quantum gravity. (As an aside, it also contradicts most mystical and spiritualistic thinking which denies the typical, naive Western over-hyping of the autonomous individual.) I remember a story by Kafka about a monkey trapped in a cage, who developed human-level intelligence with the goal of escaping the cage. I don't recall the wording but , translated into Goertzel-ese idiom, Kafka wrote something like: The monkey was not seeking freedom. By no means. Freedom is just a complicated illusion. What the monkey was seeking was something simpler and more profound and important: **a way out** ;-) This monkey is also seeking a way out, and I don't think the old illusions of free will are necessary (or sufficient) for this purpose... -- Ben G - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] What would motivate you to put work into an AGI project?
On Saturday 05 May 2007 23:29, Matt Mahoney wrote: About programming languages. I do most of my programming in C++ with a little bit of assembler. AGI needs some heavy duty number crunching. You really need assembler to do most any kind of vector processing, especially if you use a coprocessor like a graphics card or PS3 type hardware. You can get hundreds of GFlops for a few hundred dollars now, so why not use it? Look at Brook (http://graphics.stanford.edu/projects/brookgpu/) ... and GPGPU in general (http://www.gpgpu.org/cgi-bin/blosxom.cgi). If you want to use the built-in SIMD instructions in the X8x architecture, there are versions of BLAS that support them: both AMD and Intel have native versions for download, and there is ATLAS (http://math-atlas.sourceforge.net/), FFTW (http://www.fftw.org/), and many similar packages of functions -- there is also libSIMDx86 (http://sourceforge.net/projects/simdx86/) for general purpose vector and matrix processing. Josh - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Consider a ship. From one point of view, you could separate the people aboard into two groups: the captain and the crew. But another just as reasonable point of view is that captain is just one member of the crew, albeit one distinguished in some ways. One could reasonably take the point of view that the executive functions in a mind are performed by a module that is not all that much different in kind from the other ones, it just happens to be the one that is the fixpoint of the controller of relation in the architecture graph. Josh On Sunday 06 May 2007 00:18, Mike Tintner wrote: ... The human mind consists of a two-tier structure. On top, you have this conscious, executive mind that takes most of the decisions about which way the system will go - basically does the steering. On bottom, you have the unconscious, subordinate mind that does nearly all the information processing, both briefing and executing the executive mind's decisions, putting the words in its mouth and forming the thoughts in its head, while continually pressuring the executive mind with conflicting emotions, and at the same time monitoring and controlling the immensely complex operations of the body. ... You guys think you can have a successful AGI without the same basic structure? - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
As Nietzsche put it, from a functional point of view, consciousness is like the general who, after the fact, takes responsibility for the largely autonomous actions of his troops ;-) However, none of these metaphors addresses the issue of first vs. third person perspectives I hate to trumpet The Hidden Pattern again, but therein I deal with such issues at length and depth... ben On 5/6/07, J. Storrs Hall, PhD. [EMAIL PROTECTED] wrote: Consider a ship. From one point of view, you could separate the people aboard into two groups: the captain and the crew. But another just as reasonable point of view is that captain is just one member of the crew, albeit one distinguished in some ways. One could reasonably take the point of view that the executive functions in a mind are performed by a module that is not all that much different in kind from the other ones, it just happens to be the one that is the fixpoint of the controller of relation in the architecture graph. Josh On Sunday 06 May 2007 00:18, Mike Tintner wrote: ... The human mind consists of a two-tier structure. On top, you have this conscious, executive mind that takes most of the decisions about which way the system will go - basically does the steering. On bottom, you have the unconscious, subordinate mind that does nearly all the information processing, both briefing and executing the executive mind's decisions, putting the words in its mouth and forming the thoughts in its head, while continually pressuring the executive mind with conflicting emotions, and at the same time monitoring and controlling the immensely complex operations of the body. ... You guys think you can have a successful AGI without the same basic structure? - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] A New Approach to AGI: What to do and what not to do (includes my revised algorithm)
PS: [META] - people pls to cut off long message includes - some of us don't enjoy always on high bandwidth :( [META] Yes, that is a very important point for me as well. As this list is getting more and more active I'm wasting more and more time scrolling through messages (often top-posted) to find the content. Whenever you can, please cut! - lk - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
On Sunday 06 May 2007 07:49, Benjamin Goertzel wrote: As Nietzsche put it, from a functional point of view, consciousness is like the general who, after the fact, takes responsibility for the largely autonomous actions of his troops ;-) That's actually pretty close to the way (I think) it really works ... I hate to trumpet The Hidden Pattern again, but therein I deal with such issues at length and depth... As long as the trumpets are blaring, Beyond AI is coming out this month, with the coolest cover I've seen on any non-fiction book (he says modestly): http://www.amazon.com/Beyond-AI-Creating-Conscience-Machine/dp/1591025117 or just search for Beyond AI. Josh - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Consider a ship. From one point of view, you could separate the people aboard into two groups: the captain and the crew. But another just as reasonable point of view is that captain is just one member of the crew, albeit one distinguished in some ways. Really? Bush? Browne [BP, just dismissed]? Trump? Ballmer? Gates? Kapor? Semel? Branson? Sarkozy? Blair? JUST members of the crew? - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Mike, Since you mentioned me and NARS, I feel the need to clarify my position on the related issues. *. I agree with you that in many situations, the decision-making procedure doesn't follow predetermined algorithm, which give people the feeling of free will. On the other hand, at a deeper level, each basic operations in the process does roughly follow a fixed routine, and how these operations form the decision-making procedure are determined by many factors at the moment. This mechanism is already implemented in NARS, and is discussed in detail in http://nars.wang.googlepages.com/wang.computation.pdf . Whether such a process is free or determined to a large extent depends on the context of the discussion: determined by whom? given what? The system does have a choice among options from time to time, though given the design and the experience of the system, these choices are not arbitrary at all. *. I disagree with you on the two-tier structure, though it is indeed intuitively obvious. As Ben said On some topics, naive intuition can be misleading, which has been shown in many times in the history of AI and CogSci. The conscious/unconscious distinction does exist, but to me, it shows that our self-perception has its limits, just like our perception of the outside environment. I don't see your evidence for the two to be separate, rather than just different. What is your evidence for The unconscious mind thinks more or less algorithmically? To me, it is just the opposite --- to follow an algorithm needs conscious effort. If you are talking about automated behaviors or acquired skills, then that is a different issue from unconscious thinking. *. I also feel that you mixed several different issues all together in the discussion: free-will/determinism, conscious/unconscious, centralize/decentralize, which may be taken as confused philosophical understanding on your side. ;-) Pei On 5/6/07, Mike Tintner [EMAIL PROTECTED] wrote: Well, there obviously IS a conscious, executive mind, separate from the unconscious mind, whatever the enormous difficulties cognitive sicentists had in first admitting its existence and now in identifying its correlates! And you still seem to be sharing some of those old difficulties in talking about it. Science generally still has some of those difficulties too. They shouldn't be there. Social organizations have chief executives and appear more or less incapable of functioning without them. The individual organization that is a human being appears to need an executive mind for much the same reasons - though those reasons need defining. Note that Fodor acknowledges the embarrassing truth that sicence can currently offer no explanation of why the conscious mind exists - rational, deterministic computers and machines clearly do not have or need one, functioning perfectly as entirely unconscious affairs. One immediate reason, applicable to AGI - although it will take the next Cognitive Revolution to recognize this - is that the two minds, almost certainly, think very differently. The unconscious mind thinks more or less algorithmically, (at least most of the time), rapidly in set ways - like a rational computer - it has to. Its function is to get things done. The conscious mind thinks literally, freely. How long it will spend on any given decision, and what course of thought it will pursue in reaching that decision are definitely NOT set, but free. (How does Pei's NARS fit in here?) Should I buy the marshmallow or the creme caramel ice cream? Hmm that's a tough one. I want to get this right... And I could and will resolve that decision in a few more seconds OR at other times, I could still be here thinking about it several minutes later OR at other times I could wander off in mid-thought to another subject entirely. No computer currently thinks like this - thinks freely and crazily as opposed to rationally and deterministically. Anyone who produces one - that has a similar practicality to the animal/human executive mind - will literally usher in the next Cognitive Revolution. You guys are clearly moving that way - but still appear to have a somewhat confused philosophical understanding of why all this is really necessary. (One interesting, but tangential issue is that the unconscious mind does appear to have a certain freedom too - it's hard to see dreams, for example, as deterministic affairs, Well, your dreams maybe, but not mine, you understand...). - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
I find that freedom is one of those folk-psychology/philosophy concepts that isn't really much use for scientific and engineering thinking about either human or machine intelligence... As for concentration, this gets into what I call attention allocation -- an area we've paid a lot of attention to in the Novamente design. I believe an AGI should be able to adaptively combine the concentrative focus of current specialized software programs with the creativity-inducing, associatively and contextually digressive nature of human attention. -- Ben On 5/6/07, Mike Tintner [EMAIL PROTECTED] wrote: Ben, Yes, I'll match my understanding and knowledge of, and ideas on, the free will issue against anyone's. For example - and this is the real issue that concerns YOU and AGI - I just introduced an entirely new dimension to the free will debate. You literally won't find it anywhere. Including Dennett. Free thinking. If we are free to decide, then it follows we are also free to think - not merely to decide either way at the end of solving a problem, but free as to how we go about solving that problem - free to spend a little more time or less time on it, free to ask someone else's opinion or go with our gut instinct, free to list the pro's and cons or to take the first reasonable idea that comes along, free to attack it logically/algebraically or verbally etc. etc. That is an extremely important dimension of free will. It simply hasn't been considered. Clearly it should be. For the purposes of AGI, you can put the free will issue to one side, at least for a while, I would suggest, and concentrate on freedom of thought. You see, it is absolutely fundamental to robotics to describe robots in terms of degrees of freedom - of movement, (whatever your views on free will)..It is, or will be, similarly fundamental to AGI to describe autonomous computational minds in terms of degrees of freedom - of thought. There is a crashingly obvious difference between a rational computer and a human mind - and the only way cognitive science has managed not to see it is by resolutely refusing to look at it, just as it resolutely refused to look at the conscious mind in the first place. The normal computer has no problems concentrating. Give it a problem and it will proceed to produce a perfect rational train of thought, with every step taken, and not a single step missed. (Or to put that another way - it has zero freedom of thought). But human minds have major problems concentrating. Literally for more than seconds on end. For a human mind to produce a rational reflective train of thought for something like a minute is virtually impossible. Obviously this varies according to the problem/ subject, but the basic problem of concentration is acknowledged by a whole variety of psychologists from Williiam James to Cszikszentmilhalyi - and undeniable. Look at how human minds actually approach problems - their literal streams of thought (something cognitive psychology still almost totally refuses to do) - and you will find that humans can and do miss out at different times each and every step of what might be considered a rational train of thought - they don't listen to, or set the question/problem, don't look at the evidence or look at irrelevant things, don't even try to have ideas, are biassed, don't think for themselves but copy others' ideas, lose the thread, go off at tangents, repeat themelves, are uncritical, don't check etc etc. In innumerable ways, we almost always jump to conclusions and leave out ideal steps of reasoning. We are incapable of producing extended rational trains of thought and movement. (Just look at student essays, right?) We may be fairly effective reasoners, all things considered, but by the reasoning standards of rational computers we are irrational, period. Now to the rational philosopher and scientist and to the classical AI person, this is all terrible (as well as flatly contradicting one of the most fundamental assumptions of cognitive science, i.e. that humans think rationally). We are indeed only human not [rational, deterministic] machines. But I would expect someone who cares about AGI to understand that this is also all beautiful. Our extreme capacity for error can also be described as extreme freedom of thought- and the basis of our adaptivity. Every error in one context is an adaptive advantage in another. It's good and vital in all kinds of situations to be able to jump to conclusions, for example. It's good and vital to be able to completely restructure the ways you think about a problem. I would expect you and Pei to be deeply interested in that whole dimension of freedom of thought (and also to see that it provides a functional distinction between the conscious and unconscious mind, where currently NONE exists). If you are not interested, no problem. P.S. Re the free will issue, laws of physics etc, I would suggest that there is only one thing that should immediately concern
Re: [agi] The Advantages of a Conscious Mind
J Storrs Hall, PhD. writes: As long as the trumpets are blaring, Beyond AI is coming out this month, with the coolest cover I've seen on any non-fiction book (he says modestly): http://www.amazon.com/Beyond-AI-Creating-Conscience-Machine/dp/1591025117 Cool! I just pre-ordered my copy! Look at Brook (http://graphics.stanford.edu/projects/brookgpu/) ... and GPGPU in general (http://www.gpgpu.org/cgi-bin/blosxom.cgi). I'm also just beginning my experimentation with modern hardware, and just got a new machine with two Nvidia 8800GTX boards. That G80 architecture is moving explicitly to a GPGPU architecture (by which I mean it doesn't have separate vertex and pixel processors, just 128 general-purpose processors per card. They have some pretty decent programming tools for it (called CUDA). If you want to use the built-in SIMD instructions in the X8x architecture, there are versions of BLAS that support them: both AMD and Intel have native versions for download If you are working in a somewhat low-level language and don't mind a little bit of effort, you can embed the assembly directly to use the scalar functions. To get my feet wet with this, I just wrote a mandelbrot set exploration program that does this and it's amazing how far things have come recently. The CPU on my new machine is an intel quad core at 2.7 ghz. With each one executing a 4-wide simd instruction (single precision), that adds up to 43 gflops peak, which isn't anywhere near the peak of the graphics cards but isn't too shabby. As I just start to work on some AGI-type stuff myself, one of my premises is that it pays to think about models that lend themselves to efficient implementation on available hardware, in direct opposition to YKY's recent post on that subject. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Mike, Bit of confusion here. Consciousness is best used to refer to the thing that Chalmers refers to as the Hard Problem issues. The thing you are mainly referring to is what cog psych people would talk about as executive processing (as opposed to automatic processing). Big literature on that. The important thing, for me, is that I would even begin to engage you in debate on the ideas you have raised here, because it is just too messy of these two totally different ideas are mixed up together. Richard Loosemore Mike Tintner wrote: Well, there obviously IS a conscious, executive mind, separate from the unconscious mind, whatever the enormous difficulties cognitive sicentists had in first admitting its existence and now in identifying its correlates! And you still seem to be sharing some of those old difficulties in talking about it. Science generally still has some of those difficulties too. They shouldn't be there. Social organizations have chief executives and appear more or less incapable of functioning without them. The individual organization that is a human being appears to need an executive mind for much the same reasons - though those reasons need defining. Note that Fodor acknowledges the embarrassing truth that sicence can currently offer no explanation of why the conscious mind exists - rational, deterministic computers and machines clearly do not have or need one, functioning perfectly as entirely unconscious affairs. One immediate reason, applicable to AGI - although it will take the next Cognitive Revolution to recognize this - is that the two minds, almost certainly, think very differently. The unconscious mind thinks more or less algorithmically, (at least most of the time), rapidly in set ways - like a rational computer - it has to. Its function is to get things done. The conscious mind thinks literally, freely. How long it will spend on any given decision, and what course of thought it will pursue in reaching that decision are definitely NOT set, but free. (How does Pei's NARS fit in here?) Should I buy the marshmallow or the creme caramel ice cream? Hmm that's a tough one. I want to get this right... And I could and will resolve that decision in a few more seconds OR at other times, I could still be here thinking about it several minutes later OR at other times I could wander off in mid-thought to another subject entirely. No computer currently thinks like this - thinks freely and crazily as opposed to rationally and deterministically. Anyone who produces one - that has a similar practicality to the animal/human executive mind - will literally usher in the next Cognitive Revolution. You guys are clearly moving that way - but still appear to have a somewhat confused philosophical understanding of why all this is really necessary. (One interesting, but tangential issue is that the unconscious mind does appear to have a certain freedom too - it's hard to see dreams, for example, as deterministic affairs, Well, your dreams maybe, but not mine, you understand...). - Original Message - *From:* Benjamin Goertzel mailto:[EMAIL PROTECTED] *To:* agi@v2.listbox.com mailto:agi@v2.listbox.com *Sent:* Sunday, May 06, 2007 10:37 AM *Subject:* Re: [agi] The Advantages of a Conscious Mind Mike, The extent to which there is a rigid distinction between these two tiers in the human brain/mind is not entirely clear. The human brain seems to have some distinct memory subsystems associated with various sorts of short term memory or working memory, but the notion of executive processing overall is IMO best thought of as a fuzzy set. Yes, there are some parts of the brain clearly shown (by fMRI and PET) to be involved with overall coordination, but the knowledge/memories associated by these brain regions is not necessarily the totality of what can occur in subjective conscious awareness. I think that the working memory and the autonomic nervous system are best viewed as two extremes, with a continuum of conscious intensity levels existing between them. For relatively recent thinking on the underpinnings of consciousnes in the human brain, check out the edited volume -- Neural Correlates of Consciousness, by Thomas Metzinger His single-author book -- Being No One is also very good, though I disagree with his take on AI at the end of the book. (he argues it would be unethical to create AGI's because it would be unethical to experiment on their half-formed, probably buggy conscious minds.) In Novamente we do have an AttentionalFocus concept which is much like what you call the conscious tier. We have chosen the term attentional focus to avoid getting into arguments related to the nature of consciousness and the first person versus third person perspectives on mind. Each item in the attentional
Re: [agi] The Advantages of a Conscious Mind
Mike Tintner wrote: There is a crashingly obvious difference between a rational computer and a human mind - and the only way cognitive science has managed not to see it is by resolutely refusing to look at it, just as it resolutely refused to look at the conscious mind in the first place. The normal computer has no problems concentrating. Give it a problem and it will proceed to produce a perfect rational train of thought, with every step taken, and not a single step missed. (Or to put that another way - it has zero freedom of thought). Completely wrong, I am afraid. This is a view of computer that is so antiquated it belongs in the early 1960's, when people were told that computers can only do what they are programmed to do, as a way to reassure them that they should not be afraid that the computers were really able to think (and were therefore a threat). You can program a computer to be deterministic, or you can program it to be non-determinstic. You choice. Some approaches to AI do indeed take an approach that would leave the machine with no choices in its reasoning paths but that is only one choice. It is certainly not my choice, or those of many others. It is important not to tar everyone with that brush. Richard Loosemore. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Mike Tintner wrote: Now to the rational philosopher and scientist and to the classical AI person, this is all terrible (as well as flatly contradicting one of the most fundamental assumptions of cognitive science, i.e. that humans think rationally). We are indeed only human not [rational, deterministic] machines. Mike, this is getting a bit much. Your statement that one of the most fundamental assumptions of cognitive science [is] that humans think rationally is complete and utter bunk. There is no possible interpretation of this claim that could make it even slightly true. Richard Loosemore. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Mike Tintner wrote: And if you're a betting man, pay attention to Dennett. He wrote about Consciousness in the early 90's, together with Crick helped make it scientifically respectable. About five years later, consciousness studies swept science and philosophy. Nonsense. Dennett's approach was scorned by many as a whitewash. He did not make it respectable if anyone did that, it was Dave Chalmers. Crick, like many other philosophy wannabes, gave an opinion on the matter that was just a big pile of evasions. Just about everyone and their mother has written a book about consciousness, most of them trash. Dennett, although a smart cookie, bit off more than he could chew on that one. I note that he did not even bother to turn up at the Tucson conference last year. I did -- and *my* theory of consciousness was the first one ever to actually explain anything ;-) ;-). (Chalmers noticed, but I don't think anyone else did). Richard Loosemore. Now he has just written about free will, and although the book was pretty bad, it was important in being arguably the first by a scientific philosopher to assert that free will is consistent with science and materialism. I'll gladly place a friendly (and you might think outrageous) bet with you that that book is similarly prescient and free will will be the new default philosophy of science within 5-10 years. In case you haven't noticed, it is actually already being widely taken in a kind of de facto, implicit rather than explicit way, as the basic philosophy of autonomous mobile robotics. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Er nonsense to you too. :} Part of my asserting myself boldly here, was to say: look, I may be a schmuck on AI but I know a lot, here ( in fact I'll stand by the rest of my claims, - although if you guys can't recognize, for example, that free thinking opens up a new dimension on free will, then there's probably no point). Consciousness Explained ... publ. 1991. Crick's statements - 1991, Sci Am article... 1992 David Chalmers.. The Conscious Mind... Amazon gives me 1998, but it may have been 1996 - when the consciousness studies wave was already starting. Dennett and Crick were way ahead of the game and Chalmers, historically. (In fact, Crick was almost certainly the crucial figure). Sure, Consciousness Explained was attacked, though still influential. My point is a historical/ sociological one - not an evaluative one. And therefore I am perfectly entitled to make my future prediction about the sociological/ scientific significance of Freedom Evolves - I could, of course, prove totally wrong. But it's a point worth considering - IF you're interested in how culture and science are changing. And note that Dennett was even historically ahead if only just, of The God Delusion, with Breaking the Spell. (Oh, and even evaluatively, Dennett, I would argue, is the leading scientfic, i.e. pro-science, philosopher in the world. Chalmers' credentials in that respect are more dubious - not that I'm endorsing Dennett by any means). - Original Message - From: Richard Loosemore [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, May 06, 2007 4:45 PM Subject: Re: [agi] The Advantages of a Conscious Mind Mike Tintner wrote: And if you're a betting man, pay attention to Dennett. He wrote about Consciousness in the early 90's, together with Crick helped make it scientifically respectable. About five years later, consciousness studies swept science and philosophy. Nonsense. Dennett's approach was scorned by many as a whitewash. He did not make it respectable if anyone did that, it was Dave Chalmers. Crick, like many other philosophy wannabes, gave an opinion on the matter that was just a big pile of evasions. Just about everyone and their mother has written a book about consciousness, most of them trash. Dennett, although a smart cookie, bit off more than he could chew on that one. I note that he did not even bother to turn up at the Tucson conference last year. I did -- and *my* theory of consciousness was the first one ever to actually explain anything ;-) ;-). (Chalmers noticed, but I don't think anyone else did). Richard Loosemore. Now he has just written about free will, and although the book was pretty bad, it was important in being arguably the first by a scientific philosopher to assert that free will is consistent with science and materialism. I'll gladly place a friendly (and you might think outrageous) bet with you that that book is similarly prescient and free will will be the new default philosophy of science within 5-10 years. In case you haven't noticed, it is actually already being widely taken in a kind of de facto, implicit rather than explicit way, as the basic philosophy of autonomous mobile robotics. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.467 / Virus Database: 269.6.4/790 - Release Date: 05/05/2007 10:34 - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Cognitive science treats humans as thinking like computers - rationally, if boundedly rationally. Which part of cognitive science treats humans as thinking irrationally, as I have described ? (There may be some misunderstandings here which hve to be ironed out, but I don't think my claim at all outrageous or less than obvious). All the social sciences treat humans as thinking rationally. It is notorious that this doesn't fit the reality - especially for example in economics. But the basic attitude is: well, it's the best model we've got. - Original Message - From: Richard Loosemore [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, May 06, 2007 4:38 PM Subject: Re: [agi] The Advantages of a Conscious Mind Mike Tintner wrote: Now to the rational philosopher and scientist and to the classical AI person, this is all terrible (as well as flatly contradicting one of the most fundamental assumptions of cognitive science, i.e. that humans think rationally). We are indeed only human not [rational, deterministic] machines. Mike, this is getting a bit much. Your statement that one of the most fundamental assumptions of cognitive science [is] that humans think rationally is complete and utter bunk. There is no possible interpretation of this claim that could make it even slightly true. Richard Loosemore. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.467 / Virus Database: 269.6.4/790 - Release Date: 05/05/2007 10:34 - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
If you are a nondeterminist - i.e. a believer in nondeterministic programming - je t'embrasse. (see my forthcoming reply to Pei). However, having being thoroughly attacked by Ai-ers including Minsky on his group, for adopting such a position - on the basis that nondeterministic programs can be emulated by deterministic Turing machines, and don't really exist etc. etc. - and also having just been criticised by Pei, who, offhand, without much knowledge of him, I thought might be sympathetic that way - I am dubious about your representation of the situation and what is/isn't antiquated. I suspect, as re Chalemers/Dennett, you are confusing YOUR beliefs (and no doubt some others' too) about the matter with the GENERAL or most widely-held beliefs. Re cognitive science and cognitive psychology, there is one simple way to crystallise the matter. I contend that the human mind's difficulties in concentrating are one of the primary, definining characteristics of how it works, and of how it is actually programmed - and this CONTRADICTS current cog sci/psych. Show me which section of cognitive science or psychology deals with this - problems of concentration in relation to the mind's programming. Or show me any section which deals with nondeterministic programming re humans. [Cog sci/psych remember, and NOT AI]. Re the situation in AI generally, and people's attitudes to deterministic/ nondeterministic programming and what you say below, please do inform me more about how different camps think. IF I have understood this right, Ben and Pei would NOT agree with the sentiments and kind of atittude you seem to be expressing below. They don't seem to believe that freedom of thought let alone decision is possible. They would be in an opposite camp, say, to Kevin Kelly: What could be more human than to give life? I think I know: to give life and freedom. To give open-ended life. To say, here's your life and the car keys. Then you let it do what we are doing-making it all up as we go along. Tom Ray once told me, I don't want to download life into computers. I want to upload computers into life. Kevin Kelly Out of Control. The New Biology of Machines, Social Systems, and the Economic World. New York: Addison, Wesley. 1994 Kevin Kelly said to me, in an email exchange, that he reckoned that some 50% or more of AI people did believe that robots will be free. Minsky's group mocked that claim, but then they would. What do you reckon about how AI people generally stand here? - Original Message - From: Richard Loosemore [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, May 06, 2007 4:32 PM Subject: Re: [agi] The Advantages of a Conscious Mind Mike Tintner wrote: There is a crashingly obvious difference between a rational computer and a human mind - and the only way cognitive science has managed not to see it is by resolutely refusing to look at it, just as it resolutely refused to look at the conscious mind in the first place. The normal computer has no problems concentrating. Give it a problem and it will proceed to produce a perfect rational train of thought, with every step taken, and not a single step missed. (Or to put that another way - it has zero freedom of thought). Completely wrong, I am afraid. This is a view of computer that is so antiquated it belongs in the early 1960's, when people were told that computers can only do what they are programmed to do, as a way to reassure them that they should not be afraid that the computers were really able to think (and were therefore a threat). You can program a computer to be deterministic, or you can program it to be non-determinstic. You choice. Some approaches to AI do indeed take an approach that would leave the machine with no choices in its reasoning paths but that is only one choice. It is certainly not my choice, or those of many others. It is important not to tar everyone with that brush. Richard Loosemore. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.467 / Virus Database: 269.6.4/790 - Release Date: 05/05/2007 10:34 - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
On 5/6/07, Mark Waser [EMAIL PROTECTED] wrote: Yes, I'll match my understanding and knowledge of, and ideas on, the free will issue against anyone's. Arrogant much? I just introduced an entirely new dimension to the free will debate. You literally won't find it anywhere. Including Dennett. Free thinking. If we are free to decide, then it follows we are also free to think Oh, please . . . . Seriously. The only other identity I have ever encountered with such zealous believe in their own accomplishments is A. T. Murray / Mentifex. I wonder what would happen if these two super-egos (pun intended) were to collide? Sorry to contribute so little to the actual discussion, but really... - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Pei, Thanks for stating your position (which I simply didn't know about before - NARS just looked at a glance as if it MIGHT be nondeterministic). Basically, and very briefly, my position is that any AGI that is to deal with problematic decisions, where there is no right answer, will have to be freely, nondeterministically programmed to proceed on a trial and error basis - and that is just how human beings are programmed. (Nondeterministically programmed should not be simply equated with current kinds of programming - there are an infinity of possible ways of programming deterministically, ditto for nondeterministically). Some of what you say below IS confusing - The system does have a choice among options from time to time, though given the design and the experience of the system, these choices are not arbitrary at all. That sounds like a complete contradiction in terms. Either you have a real choice or not. Let's say the system is investing in the stockmarket - if it's free, in my terms, it will indeed have a choice, and be able to Buy, OR Sell OR Hold. If it's determined, or not arbitrary at all, it will at a given point, have only ONE option open to it. Can you clarify your position? I'm somewhat confused too by: What is your evidence for The unconscious mind thinks more or less algorithmically? To me, it is just the opposite --- to follow an algorithm needs conscious effort. My position is this: most of our behaviour is unconsciously controlled. When you walk across a room, most steps will be automatic. When I wrote that last sentence most if not all of the words and letters and keypresses were automatic. And I assume there are unconscious algorithms/ routines controlling those behaviours But while most of our steps on any given journey are automatic and fixed, we also more or less continuously consciously and deliberately and freely attend to the occasional next step and turn - and how, and how long we think about and take that next step is not fixed. [So if you are going to argue that it's not algorithms but some other kind of deterministic programming that does the unconscious controlling, I wouldn't try and argue about that} What I find weird is your statement - an algorithm needs conscious effort. Then it's not an algorithm, or any kind of deterministic programming. Nothing that requires conscious exertion can be algorithmic or deterministic or automatic. Effort/exertion - i.e. whether to make it or not - is fundamentally problematic and nondeterministic.When you are doing your fiftieth or maximal press-up, there is no algorithm or any oither kind of deterministic programming that determines whether you will push beyond your limit to the fifty-fifth. You face a problematic decision as to whether you are or are not prepared to make the exertion and bear the pain of higher achievement or stop now and settle for less achievement with less pain. When you are straining sexually, and agonizing over whether to keep going, there is no algorithm that determines whether you will keep bearing the tension for another thirty seconds, or one minute or whatever. You have a problematic decision as whether you are prepared to aim for still more pleasure AND still more pain, or come now and settle for less pleasure and less pain - and there is no right answer.. Daniel knows that Allison needs at least another five minutes of intercourse before she can climax. Here's the problem: Daniel doesn't think he has five minutes left in him. If Daniel continues having intercourse the way he has for the past ten minutes, it may be only a matter of seconds before he has an orgasm. He thinks about slowing down or stopping. Besides, if he tried to stop or to change the rhythm, Daniel could lose strength in his erection, which would complicate matters even further. This dilemma is making the whole experience a lot less pleasurable for Daniel. Barbra Keesling, How To Make Love All Night (And Drive A Woman Wild). 1994 Daniel here is not controlled by any deterministic algorithm or programming. Do you really - hand on heart and hope to die - believe he is? You will note that the concepts of struggle, exertion, nerve, grit etc are more or less entirely missing from scientific psychology. They are simply incompatible with a deterministic approach to the human mind, so science does what it always does in such situations - ignores them. Science doesn't deal with Daniel's problem, but in one form or other, AGI, I believe, will have to. - Original Message - From: Pei Wang [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, May 06, 2007 3:47 PM Subject: Re: [agi] The Advantages of a Conscious Mind Mike, Since you mentioned me and NARS, I feel the need to clarify my position on the related issues. *. I agree with you that in many situations, the decision-making procedure doesn't follow predetermined algorithm, which give people the
Re: [agi] rule-based NL system
Well I will go with the high level of intelligence condition, and I would think it is pretty obvious. We know already that among humans there is a grading or levels of intelligence, so unless there is some specific thing you must have to be intelligent, I would consider a 20 yr old, a 10, and a 5 yr old intelligent, and measure the intelligence with a list of things they can do, they can walk, talk move around blocks, etc, the extent they can accomplish what they want. A quadrapeligic who cant move but can only type is still intelligent, What about a brain damaged person with alzeihmers? They cant remember well but maybe they can still dress and eat by themselves, just not hold a job. A savant that can be trained to water the flowers in a garden? eh cant do anything else btu this one function, but he can look and tell if they need water, and which ones to water, and can accept instruction.. I think that is still intelligentn behavior, but is extremely limited. Dogs can be trained to rescue or so search out drugs, which is intelligent, but a narrow usage. Expert systems are quite smart in their domains, and thermostats have a range of intelligence. Ours here at the house has one box upstairs and downstiars controlled by a main unit, that could do a range of things. High-level or approaching human level intelligence is what most of us are all concerned with here, but I think in defining intelligence we have to be able to look all the way up and down the range that it offers and recognize these as having intelligence. If you dont call a thermostat intelligent, then you have to in some other way define what it does, either by saying its an object that makes decisions based on input or simply programmed or whatnot, these all boil down and start looking like our various intelligence definitions, accept input, make decisions, give output, try to reach a goal Anything lacking one of those 4 components I might not think of as intelligent. James Ratcliff Mark Waser [EMAIL PROTECTED] wrote: My view of intelligence is rather different. I don't believe that a thermostat has intelligence (and saying so tends to invite ridicule which is bad public relations). I *do* understand your point but saying that a thermostat has intelligence violates the common man's understanding of intelligence -- and that is not a good thing to do unless you have very good reason. Maybe you should just assume that my intelligence is equivalent to your high-level of intelligence. If you're willing to do so, though, I'll immediately ask why you need to call a non-high-level of intelligence intelligent.:-) Mark - Original Message - From:James Ratcliff To: agi@v2.listbox.com Sent: Saturday, May 05, 2007 1:33 AM Subject: Re: [agi] rule-based NLsystem Its mainly that Ibelieve there is a full range of intelligences available, from a simplethermostat, to a complex one that measures and controls humudity and knows ifa person is in a run, and has specific settings for differnt people, to a anexpert system, to a human to an AI and super AGI, all having some level ofintelligence. The ones we are concerned with are the 1/2 human leveland anything above. Learning I would say is a key role inhaving a high-level of intelligence, probably the main building block,learning and reasoning, both tied tightly together. James Ratcliff Mark Waser[EMAIL PROTECTED] wrote: I would say rote memorization and knowledge / data, IS understanding. OK, we have a definitional difference then. My justification for my view is that I believe that you only *really* understand something when you have predictive power on cases that you haven't directly seen yet (sort of like saying that, in order to be useful or have any value, a hypothesis must have predictive power). I look outside and I see a tree, I understand that it is a tree, I know its a tree, I know about leaves and grass and how it grows... I havnt learned anything new, I memorized all that from books and teaching etc. I don't think so. I think that you have a lot of information that you derived from generalizations, analogies, etc (i.e. learning). I would further say that I given the level of knowledge and understanding about the tree that I was intelligent in that area, you could ask me questions and I could answer them, I could conjecture what would happen if I dug the tree up etc. Are you *sure* that you've been directly told what would happen if you dug a tree up? What do you think would happen if you dug up a planticus imaginus? I'm sure that you haven't been specifically told what would happen then. :-) I think that you have some
Re: [agi] The Advantages of a Conscious Mind
Without getting into what consciousness is in humans, and how that works, some type of controller or attention module must be done in an AGI, because given a wide range of options and goals, it must allocate its time and enery into what it should be doign at any one point in time. The design of this single module woudl be very interesting to look at. A simple case is physically watching a scene, and attention is grabbed whenever motion is seen, such as a car passing by you or a bird flying past the window. What will control the attention of an AGI though? It is preumably progrqammed to accept input and directions from us, but it must have a Motivational module to make decisions about what is important as well. I dont think there is anything mystical about free-will / consiousness when applied to AGI though. On some level the AGI will have some form of autonomy, if nothing else, then at a low decicision making choice it will have the ability to say, I choose A over B randomly when no other factors are involved. What level of autonomy and how much freedom they have will be an intersting thing to follow. James Ratcliff Mike Tintner [EMAIL PROTECTED] wrote: YKY: Consciousness is not central to AGI . The human mind consists of a two-tier structure. On top, you have this conscious, executive mind that takes most of the decisions about which way the system will go - basically does the steering. On bottom, you have the unconscious, subordinate mind that does nearly all the information processing, both briefing and executing the executive mind's decisions, putting the words in its mouth and forming the thoughts in its head, while continually pressuring the executive mind with conflicting emotions, and at the same time monitoring and controlling the immensely complex operations of the body. (Forget about consciousness/ sentience here - the big deal is simply that two-tier structure). You guys think you can have a successful AGI without the same basic structure? - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; ___ James Ratcliff - http://falazar.com Looking for something... - Looking for earth-friendly autos? Browse Top Cars by Green Rating at Yahoo! Autos' Green Center. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Mike, I believe many of the confusions on this topic is caused by the following self-evident belief: A system is fundamentally either deterministic or non-deterministic. The human mind, with free will, is fundamentally non-deterministic; a conventional computer, being Turing Machine, is fundamentally deterministic. Based on such a belief, many people think AGI can only be realized by something that is non-deterministic by nature, whatever that means. This belief, though works fine in some other context, is an oversimplification in the AI/CogSci context. Here, as I said before, whether a system is deterministic may not be taken as an intrinsic nature of the system, but as depending on the description about it. For example, NARS is indeed nondeterministic in the usual sense, that is, after the system has obtained a complicated experience, it will be practically impossible for either an observer or the system itself to accurately predict how the system will handle a user-provided task. On the other level of description, NARS is still a deterministic Turing Machine, in the sense that its state change is fully determined by its initial state and its experience, step by step. Now the important point is: when we say that the mind is nondeterministic, in what sense are we using the term? I believe it is like it will be practically impossible for either an observer or the mind itself to accurately predict how the system will handle a problem, rather than it will be theoretically impossible for an observer to accurately predict how the system will handle a problem, even if the observer has full information about the system's initial state, processing mechanism, and detailed experience, as well as has unlimited information processing power. Therefore, for all practical considerations, including the ones you mentioned, NARS is nondeterministic, since it doesn't process input tasks according to a task-specific algorithm. [If the above description still sounds confusing or contradictionary, you'll have to read my relevant publications. I don't have the intelligence to explain everything by email.] Pei On 5/6/07, Mike Tintner [EMAIL PROTECTED] wrote: Pei, Thanks for stating your position (which I simply didn't know about before - NARS just looked at a glance as if it MIGHT be nondeterministic). Basically, and very briefly, my position is that any AGI that is to deal with problematic decisions, where there is no right answer, will have to be freely, nondeterministically programmed to proceed on a trial and error basis - and that is just how human beings are programmed. (Nondeterministically programmed should not be simply equated with current kinds of programming - there are an infinity of possible ways of programming deterministically, ditto for nondeterministically). - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] What would motivate you to put work into an AGI project?
One goal or project I was considering (for profit) is a research tool, basically a KB that scans in teh newspapers and articles and extracts pertinent information for others to query against and use. This would help build up a large world knowledge base, and would also be salable to research companies and such. One example of that is the tragedy shooting at VT this past month, I ran some scritps against the news article and came up with a lot of hidden information in there about the Chu guys family and some other conenctions that I wasnt seeing in many of the news articles, that let me go down some other paths to find info. Another goal or application was a 3D avatar bot like Novamente is now pursuing. This could be used most easily to simulate an autonomous AGI agent that could act in a 3d rich world. James Ratcliff Matt Mahoney [EMAIL PROTECTED] wrote: About business. Do you have any specific project goals? Something that might bring in money in the next 3-5 years? It is OK with me if our goal is to build something and give it away. A lot of people have made money that way. Look at Linux. I gave away my PAQ compressor and I've gotten 3 consulting jobs as a result, not counting work I turned down, and I never even looked for work. I just don't want to make the same mistake as Cyc and build something that nobody can use. I know AGI has lots of potential applications, but how are we going to show that our AGI is better than our competition? --- YKY (Yan King Yin) wrote: Hi =) I already have a project going on.. but it's still in the planning stage. The main difficulty is finding people who agree in the main about the basic theory. About my project: 1. Has to be for-profit, but openness is good. Also it'd be quite different from conventional companies in that the project is owned by all partners and decisions are made by voting. 2. Knowledge representation is basically FOPL, perhaps with probabilities / fuzziness. This rules out scruffie AI folks, sorry. Everyone knows that intelligence entails a lot of things (eg vision), but I believe there should be a core that is based on a uniform representation. Guess it's better to skip the scruffie vs neat debate, and simply let people coalesce to different projects. These 2 are the most important criteria. I tend to prefer partners with a more theoretical slant, rather than churning out code at high speed. Some minor points: a) language -- unimportant. I think I'll use Lisp for initial development, then switch to probably C# or Java. It's so difficult to find the right minds that language should not be a cause of disagreement at all. The entire project doesn't need to be in same language, but I also believe that it would not be colossal in size. b) reflection -- source-level reflection is not needed for a basically declarative AGI. Note that this doesn't mean my AGI would not be able to program itself eventually. c) well-documented, sure. d) chat room: I say let's start a chat room for AGI in general. I have started one on freenode.net, channel = #General-Intelligence (for some reason the names #AGI and #GI were taken). e) I'd like to be able to say everyone can do their own thing but there should be some structure that people can agree to, which I think is the KR. Cheers! ___ James Ratcliff - http://falazar.com Looking for something... - Ahhh...imagining that irresistible new car smell? Check outnew cars at Yahoo! Autos. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Mike Tintner wrote: Cognitive science treats humans as thinking like computers - rationally, if boundedly rationally. Which part of cognitive science treats humans as thinking irrationally, as I have described ? (There may be some misunderstandings here which hve to be ironed out, but I don't think my claim at all outrageous or less than obvious). All the social sciences treat humans as thinking rationally. It is notorious that this doesn't fit the reality - especially for example in economics. But the basic attitude is: well, it's the best model we've got. It is hard to argue with you when you make statements that so flagrantly contradict the facts: pick up a textbook of cognitive psychology (my favorite is Eysenck and Keane, but you can try John Anderson...) and you will find some chapters that specifically discuss the experimental evidence for the fact that humans do not generally think in rational ways. They study the irrationality, so how could they possibly assume that humans are rational like computers? These people would not for one minute go along with your statement that they assume that humans think like computers. That term rational is crucial. I am using it the way everyone in cognitive science uses it. Which part of cognitive science treats humans as thinking irrationally? Egads: all of it! Richard Loosemore. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Mike Tintner wrote: Er nonsense to you too. :} Part of my asserting myself boldly here, was to say: look, I may be a schmuck on AI but I know a lot, here ( in fact I'll stand by the rest of my claims, - although if you guys can't recognize, for example, that free thinking opens up a new dimension on free will, then there's probably no point). Consciousness Explained ... publ. 1991. Crick's statements - 1991, Sci Am article... 1992 David Chalmers.. The Conscious Mind... Amazon gives me 1998, but it may have been 1996 - when the consciousness studies wave was already starting. Dennett and Crick were way ahead of the game and Chalmers, historically. (In fact, Crick was almost certainly the crucial figure). Sure, Consciousness Explained was attacked, though still influential. My point is a historical/ sociological one - not an evaluative one. And therefore I am perfectly entitled to make my future prediction about the sociological/ scientific significance of Freedom Evolves - I could, of course, prove totally wrong. But it's a point worth considering - IF you're interested in how culture and science are changing. And note that Dennett was even historically ahead if only just, of The God Delusion, with Breaking the Spell. (Oh, and even evaluatively, Dennett, I would argue, is the leading scientfic, i.e. pro-science, philosopher in the world. Chalmers' credentials in that respect are more dubious - not that I'm endorsing Dennett by any means). I have no interest in what dates people came out with their books, I am only interested in the content of their ideas and the influence they have had on the research community. Dennett produced a muddle. Crick came out with an idea that tried to look scientific but was a sham. Chalmers, for all his faults, shed a clarifying light on the whole situation and has been justly lauded for having done so. By writing what he did, he put Dennett and Crick in perspective. But these philosophy debates can get even more exhausting than AGI ones: I am happy to accept that you have a different opinion on the matter, and leave it at that. Richard Loosemore. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] rule-based NL system
What about a brain damaged person with alzeihmers? At the risk of being politically incorrect, on a bad day -- pretty much unintelligent (though still capable to some degree) A savant that can be trained to water the flowers in a garden? eh cant do anything else btu this one function, but he can look and tell if they need water, and which ones to water, and can accept instruction.. I think that is still intelligentn behavior, but is extremely limited. Exactly as you say. Intelligent -- but limited intelligence. Dogs can be trained to rescue or so search out drugs, which is intelligent, but a narrow usage. OK. Expert systems are quite smart in their domains, But, unless they learn, not intelligent. and thermostats have a range of intelligence. Nope. They can't learn. High-level or approaching human level intelligence is what most of us are all concerned with here, but I think in defining intelligence we have to be able to look all the way up and down the range that it offers and recognize these as having intelligence. I cut off the range with learning. It's not clear to me where you cut off the range but if you include thermostats, I think you're going too far.:-) If you don't call a thermostat intelligent, then you have to in some other way define what it does, either by saying its an object that makes decisions based on input or simply programmed or whatnot, these all boil down and start looking like our various intelligence definitions, accept input, make decisions, give output, try to reach a goal All of your definitions for the thermometer are fine but since my definition of intelligence says speed of learning and it doesn't learn, it ain't intelligent. Anything lacking one of those 4 components I might not think of as intelligent. Except that I make it 5 components (and that last component -- learning -- pretty much sums up the difference between our definitions). - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
On Sunday 06 May 2007 10:18, Mike Tintner wrote: Consider a ship. From one point of view, you could separate the people aboard into two groups: the captain and the crew. But another just as reasonable point of view is that captain is just one member of the crew, albeit one distinguished in some ways. Really? Bush? Browne [BP, just dismissed]? Trump? Ballmer? Gates? Kapor? Semel? Branson? Sarkozy? Blair? JUST members of the crew? Your point being, I assume, that the executive module doesn't even have to have as much intelligence as the average member module... Josh - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
On Sunday 06 May 2007 09:47, Mike Tintner wrote: And if you're a betting man, pay attention to Dennett. He wrote about Consciousness in the early 90's, together with Crick helped make it scientifically respectable. Actually, the serious study of consciousness was made respectable by Julian Jaynes in '76 with the publication of Origin of Consciousness in the Breakdown of the Bicameral Mind. Psychologists at Rutgers I discussed it with at the time assured me that Jaynes had rock-solid credentials (he was at Princeton at the time), and so that even though nobody thought the theory was right, there was a sea-change away from thinking it was silly to theorize about at all. Note that Libet's famous work was mostly published in the early 80's. Josh - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
On Sunday 06 May 2007 09:47, Mike Tintner wrote: For example - and this is the real issue that concerns YOU and AGI - I just introduced an entirely new dimension to the free will debate. Everybody and his dog, especially the philosophers, thinks that they have some special insight into free will, and frankly they're all hooey, especially the philosophers. The only person, for my money, who has really seen through it is Drew McDermott, Yale CS prof (former student of Minsky). He points out that almost any straightforward mental architecture for a robot that models the world for planning purposes will perforce model itself as being excluded from the determinism of the rest of the model. The whole theory fits on a page and you can read it in McDermott's book (Mind and Mechanism) or my rendition in Beyond AI. In my humble opinion, McDermott has demolished 3 millenia of philosophical mumbo-jumbo, and now that we understand what free will actually means in a mental architecture, we should set about the business of implementing it. Josh Ps -- this won't stop the philosophers, of course. They would refer to DM's explanation as an error theory, namely one describing why people think they have free will instead of saying what it really is. They can then happily spend the next 3 millenia telling our AIs that they don't have real free will, though the AIs will have an unshakable intuition that they do (just like us). - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Picture Tree
Richard, I don't think I'm not getting it at all. What you have here is a lot of good questions about how the graphics level of processing that I am proposing, might work. And I don't have the answers, and haven't really thought about them yet. What I have proposed is a general idea loosely outlining 3 levels of processing. Now if it's right, that alone is valuable. And there is at least some evidence to think it might be - starting with the strange fact that blind people produce graphics drawings, and the abundance of graphics sign systems, plus, although I didn't really deal with this, that humans do have difficulties understanding abstract verbal statements. I obviously haven't had time to demonstrate it to you,. but the idea does start to impose order on our sign systems and it's quite hard just to do that. All scientific ideas and theories only go so far and spell out things in limited detail. The fact that they are not more detailed is NOT per se an objection to them. People objected to Newton - [no, I am NOT comparing this idea in any way with his work] - because he didn't spell out how gravity worked. He didn't have to. What he showed about gravitational attraction was enough. I can't see that ANY of your questions pose an immense brick wall. If you were able to argue, for argument's sake: look, the human brain simply can't handle graphics outlines, only symbolic formulae that WOULD be a brick wall. Just consider your questions again. If I ask you, for example, to visualise a graphic of a man, and a penny you will, I suggest, do it. Your brain WILL produce relevant graphics. Now how did it do that? Why did it pick those particular graphics, given that you have vast numbers available to you? Hey, neither you nor I have the answer to how it did that (although we can think about it another time). But on one level, what does it matter? The point is: IT DID IT Your brain was not stymied, as your questions seem to imply it should be; it just went ahead. Similarly, how does the brain achieve visual object recognition? How does it manage to recognize cats and dogs? What templates does it use? How does it manage to select a particular cat template, when it may well have hundreds? I think we can be confident that it does use a template or templates one way or another. Perhaps it just grabs the nearest one at neuronal hand. (And BTW I'd be v.. interested to discuss all this in another thread). But, whatever, the brain does it. ... But if I were to be guided by the spirit of your objections, I would, say: hey I can think no more about this, the whole idea is ridiculous. Ditto re your objections as to how the brain could create moving graphics as I propose. No, I don't know exactly how it does it. Here's a frame from a dream of mine - a man with a beard on flame, in a check shirt, lying on the ground. I doubt that I have ever seen that bearded head, with that check shirt, lying in that posture, let alone on flame. The brain combined four new elements in a flash in a new moving picture. If it can do that, there is no reason as yet to think that it can't create moving graphics, or moving images if necessary to test out sentences as I propose. But re your mental models, I'm just asking, what on earth do you and others mean? I'm sure whatever you're proposing is possible, I'd just like to know what it is - and if I'm confused, i.e. find the whole concept vague, then I'm pretty confident you and everyone else are also confused - because, according to my theory, (and I believe this is true), the brain DEMANDS to have concepts like that make sense. It complains - positively aches to a greater or lesser degree - if they don't, and yours will have already. (Repeat: there is no a priori objection to the concept of mental models). There is an irony - you have just asked fifteen or so questions of my graphics idea, and not one of mental models. P.S. A personal comment here - it's offered as an intuitive response, not a reasoned judgment, if it's no use or wong, screw it. You have just offered an awful lot of what are actuially constructive suggestions and proposals for further thought, as if they were damning objections. I felt intuitively that you were dong something similar in trying to define intelligence - trying to take things to minute pieces - and in the end, turning an initially constructive drive into a negative conclusion. Like I said, a purely intuitive response, and my apologies if it's wrong or no use. There is ONE BIG THING HERE THAT YOU ARE NOT GETTING. If you were to sit down and try to implement an actual system that did the above, how would you get it actually DO the drawing? What mechanisms would be in there that, after looking at the WORDS, would conclude from the words the man climbed the penny that a drawing of a penny and a man were involved? How would those mechanisms choose what kind of man, what kind of penny,
Re: [agi] The Advantages of a Conscious Mind
Pei, I don't think there's any confusion here. Your system as you describe it IS deterministic. Whether an observer might be confused by it is irrelevant. Equally the fact that it is determined by a complex set of algorithms applying to various tasks and domains and not by one task-specific algorithm, is also irrelevant. It's still deterministic. The point, presumably, is that your system has a clear set of priorities in deciding between different goals, tasks, axioms and algorithms Humans don't. Humans are still trying to work out what they really want, and what their priorities are between, for example, the different activities of their life, between work, sex, friendship, love, family etc. etc. Humans are designed to be in conflict about their fundamental goals throughout their lives. And that, I would contend, is GOOD design, and essential for their success and survival. If there's any confusion, think about many women and dieting. They will be confronted by much the same decisions about whether to eat or not to eat on possibly thousands of occasions throughout their lives. And over and over, throughout their entire lives, they will - freely - decide now this way, now that. Yo-yoing on and off their diets. Your system, as I understand it, would never do that - would never act in such crazy, mixed up, contradictory ways. Humans do, because they are, truly, free - and, I contend, non-deterministically programmed - and, repeat, this is, paradoxically, good design.. - Original Message - From: Pei Wang [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, May 06, 2007 8:48 PM Subject: Re: [agi] The Advantages of a Conscious Mind Mike, I believe many of the confusions on this topic is caused by the following self-evident belief: A system is fundamentally either deterministic or non-deterministic. The human mind, with free will, is fundamentally non-deterministic; a conventional computer, being Turing Machine, is fundamentally deterministic. Based on such a belief, many people think AGI can only be realized by something that is non-deterministic by nature, whatever that means. This belief, though works fine in some other context, is an oversimplification in the AI/CogSci context. Here, as I said before, whether a system is deterministic may not be taken as an intrinsic nature of the system, but as depending on the description about it. For example, NARS is indeed nondeterministic in the usual sense, that is, after the system has obtained a complicated experience, it will be practically impossible for either an observer or the system itself to accurately predict how the system will handle a user-provided task. On the other level of description, NARS is still a deterministic Turing Machine, in the sense that its state change is fully determined by its initial state and its experience, step by step. Now the important point is: when we say that the mind is nondeterministic, in what sense are we using the term? I believe it is like it will be practically impossible for either an observer or the mind itself to accurately predict how the system will handle a problem, rather than it will be theoretically impossible for an observer to accurately predict how the system will handle a problem, even if the observer has full information about the system's initial state, processing mechanism, and detailed experience, as well as has unlimited information processing power. Therefore, for all practical considerations, including the ones you mentioned, NARS is nondeterministic, since it doesn't process input tasks according to a task-specific algorithm. [If the above description still sounds confusing or contradictionary, you'll have to read my relevant publications. I don't have the intelligence to explain everything by email.] Pei On 5/6/07, Mike Tintner [EMAIL PROTECTED] wrote: Pei, Thanks for stating your position (which I simply didn't know about before - NARS just looked at a glance as if it MIGHT be nondeterministic). Basically, and very briefly, my position is that any AGI that is to deal with problematic decisions, where there is no right answer, will have to be freely, nondeterministically programmed to proceed on a trial and error basis - and that is just how human beings are programmed. (Nondeterministically programmed should not be simply equated with current kinds of programming - there are an infinity of possible ways of programming deterministically, ditto for nondeterministically). - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.467 / Virus Database: 269.6.4/790 - Release Date: 05/05/2007 10:34 - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to:
Re: [agi] The Advantages of a Conscious Mind
If there's any confusion, think about many women and dieting. They will be confronted by much the same decisions about whether to eat or not to eat on possibly thousands of occasions throughout their lives. And over and over, throughout their entire lives, they will - freely - decide now this way, now that. Yo-yoing on and off their diets. Your system, as I understand it, would never do that - would never act in such crazy, mixed up, contradictory ways. Humans do, because they are, truly, free - and, I contend, non-deterministically programmed - and, repeat, this is, paradoxically, good design.. Mike, I don't want to be insulting, but you seem incredibly confused about some basic concepts. Either that or you are redefining basic words in such odd ways that communicating with you usefully is next to impossible! There is no reason at all why a deterministic system couldn't yo-yo on and off a diet. I don't understand why you would think so. There is nothing stopping deterministic systems from being confused, idiotic, self-contradictory, etc. Really. Not unless you are adopting a very very strange and nonstandard definition of deterministic. I think I am going to stop responding to your messages, personally, because we simply are not communicating in a useful way. -- Ben G - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
On May 6, 2007, at 2:27 PM, J. Storrs Hall, PhD. wrote: The only person, for my money, who has really seen through it is Drew McDermott, Yale CS prof (former student of Minsky). He points out that almost any straightforward mental architecture for a robot that models the world for planning purposes will perforce model itself as being excluded from the determinism of the rest of the model. The whole theory fits on a page and you can read it in McDermott's book (Mind and Mechanism) or my rendition in Beyond AI. In my humble opinion, McDermott has demolished 3 millenia of philosophical mumbo-jumbo, and now that we understand what free will actually means in a mental architecture, we should set about the business of implementing it. Eh? Unless McDermott first came up with that idea long before he wrote that book, it is just a rehash of a relatively old idea. It is a trivial consequence of the elementary theorems of computational information theory; the necessary mathematics to prove this basic characteristic is how my copy of Li Vitanyi introduces Chapter 2. I agree with the general argument, but unless McDermott has been making this argument a *long* time, his argument is more of a me too one AFAICT. Perhaps he put his own flavor to it, but the underlying principle is not particularly new. Cheers, J. Andrew Rogers - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Mark, Indeed. Many confusions are caused by the ambiguity and context dependency of terms in natural languages. For this reason, it is not a good idea to simply label a system as deterministic or non-deterministic without clarifying the sense of the term. Pei On 5/6/07, Mark Waser [EMAIL PROTECTED] wrote: Hi Pei, I liked your definition so I went to dictionary.com and found two different definitions of deterministic which seem to clearly show our dilemma === Free On-line Dictionary of Computing - Cite This Source deterministic 1. Describes a system whose time evolution can be predicted exactly. Contrast probabilistic. For all practical purposes, NARS and the human mind are non-deterministic by this definition. === WordNet - Cite This Source deterministic adjective an inevitable consequence of antecedent sufficient causes And I would argue that both the human mind and NARS are deterministic by this definition.:-) === Makes it kind of tough to argue, doesn't it? - Original Message - From: Pei Wang [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, May 06, 2007 3:48 PM Subject: Re: [agi] The Advantages of a Conscious Mind Mike, I believe many of the confusions on this topic is caused by the following self-evident belief: A system is fundamentally either deterministic or non-deterministic. The human mind, with free will, is fundamentally non-deterministic; a conventional computer, being Turing Machine, is fundamentally deterministic. Based on such a belief, many people think AGI can only be realized by something that is non-deterministic by nature, whatever that means. This belief, though works fine in some other context, is an oversimplification in the AI/CogSci context. Here, as I said before, whether a system is deterministic may not be taken as an intrinsic nature of the system, but as depending on the description about it. For example, NARS is indeed nondeterministic in the usual sense, that is, after the system has obtained a complicated experience, it will be practically impossible for either an observer or the system itself to accurately predict how the system will handle a user-provided task. On the other level of description, NARS is still a deterministic Turing Machine, in the sense that its state change is fully determined by its initial state and its experience, step by step. Now the important point is: when we say that the mind is nondeterministic, in what sense are we using the term? I believe it is like it will be practically impossible for either an observer or the mind itself to accurately predict how the system will handle a problem, rather than it will be theoretically impossible for an observer to accurately predict how the system will handle a problem, even if the observer has full information about the system's initial state, processing mechanism, and detailed experience, as well as has unlimited information processing power. Therefore, for all practical considerations, including the ones you mentioned, NARS is nondeterministic, since it doesn't process input tasks according to a task-specific algorithm. [If the above description still sounds confusing or contradictionary, you'll have to read my relevant publications. I don't have the intelligence to explain everything by email.] Pei On 5/6/07, Mike Tintner [EMAIL PROTECTED] wrote: Pei, Thanks for stating your position (which I simply didn't know about before - NARS just looked at a glance as if it MIGHT be nondeterministic). Basically, and very briefly, my position is that any AGI that is to deal with problematic decisions, where there is no right answer, will have to be freely, nondeterministically programmed to proceed on a trial and error basis - and that is just how human beings are programmed. (Nondeterministically programmed should not be simply equated with current kinds of programming - there are an infinity of possible ways of programming deterministically, ditto for nondeterministically). - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
On 5/6/07, Mike Tintner [EMAIL PROTECTED] wrote: Pei, I don't think there's any confusion here. Your system as you describe it IS deterministic. Whether an observer might be confused by it is irrelevant. Equally the fact that it is determined by a complex set of algorithms applying to various tasks and domains and not by one task-specific algorithm, is also irrelevant. It's still deterministic. OK, let's use the word in this way. Then how do you know that the human mind is not deterministic in this sense? Just because you don't know a complex set of algorithms that can explain its behaviors? The point, presumably, is that your system has a clear set of priorities in deciding between different goals, tasks, axioms and algorithms Wrong. NARS often needs to work hard to decide between different goals, tasks, axioms and algorithms, and is not always successful in doing that. You confused the algorithms in a system that make it work with algorithms defined with respect to problem classes. Humans don't. Humans are still trying to work out what they really want, and what their priorities are between, for example, the different activities of their life, between work, sex, friendship, love, family etc. etc. Humans are designed to be in conflict about their fundamental goals throughout their lives. And that, I would contend, is GOOD design, and essential for their success and survival. Agree, but the same description is true for NARS, in principle. If there's any confusion, think about many women and dieting. They will be confronted by much the same decisions about whether to eat or not to eat on possibly thousands of occasions throughout their lives. And over and over, throughout their entire lives, they will - freely - decide now this way, now that. Yo-yoing on and off their diets. Your system, as I understand it, would never do that - would never act in such crazy, mixed up, contradictory ways. Your understanding about NARS is completely wrong. Can you tell me which publications of mine give you this impression? Or you simple assume that all deterministic systems must behave in this way? Pei - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
On Sunday 06 May 2007 17:59, J. Andrew Rogers wrote: On May 6, 2007, at 2:27 PM, J. Storrs Hall, PhD. wrote: The only person, for my money, who has really seen through it is Drew McDermott, Yale CS prof (former student of Minsky). ... Eh? Unless McDermott first came up with that idea long before he wrote that book, it is just a rehash of a relatively old idea. ... Assuming we're thinking about the same book, Li Vitanyi was published in 1993. McDermott came up with his theory/explanation in the 80's and published it on the ARPANET AI list (which is where I first came across it). Josh - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
J. Storrs Hall, PhD. writes: I'm intending to do lo-level vision on (one) 8800 and everything else on my (dual) Clovertowns. Do you have any particular architectures / algorithms you're working on? Your approach and mine sound like there could be valuable shared effort... First I'm going to build a robot. While I do that, I'm going to learn how to use the GPU hardware, read a lot, and figure out what to do next. However, I'm definitely planning on starting with low level vision on the 8800 so we're certainly going in the same direction in that regard. So far I'm capturing video from a firewire webcam using the CMU 1394 camera driver, but haven't yet started doing much with the data except displayi it. It should be possible to run hundreds of different convolutions on image data in realtime so I'm planning to do that as a learning project. I'm curious whether a clustering algorithm would automatically come up with useful convolution kernels naturally simply by watching vast quantities of image data (somebody must have tried that at some point), but it also isn't too hard to hardcode a bunch of oriented edge detectors, endpoint detectors, corner detectors, and whatnot. I have no idea where to go from there at this point, but that's the fun of it. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
On May 6, 2007, at 4:08 PM, J. Storrs Hall, PhD. wrote: On Sunday 06 May 2007 17:59, J. Andrew Rogers wrote: On May 6, 2007, at 2:27 PM, J. Storrs Hall, PhD. wrote: The only person, for my money, who has really seen through it is Drew McDermott, Yale CS prof (former student of Minsky). ... Eh? Unless McDermott first came up with that idea long before he wrote that book, it is just a rehash of a relatively old idea. ... Assuming we're thinking about the same book, Li Vitanyi was published in 1993. McDermott came up with his theory/explanation in the 80's and published it on the ARPANET AI list (which is where I first came across it). Ah, okay, that would be a bit before my time. :-) I've been aware of similar arguments since something like the late-80s, but not from ARPANET. Proofs of the necessary theorems have been around since the mid-1960s and important ever since. I would be surprised if the idea did not pre-date the 1980s. My point about Li Vitanyi was more that it is considered elementary in the scheme of things and has been for a long time, not that it was original to that book. It surprises me that people actually in the field still find the consequences of it to be controversial. Cheers, J. Andrew Rogers - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Picture Tree
Mike, I really don't know what to say any more. Too much of what you suggest has been considered in great depth by other people. It is an insult to them, if you ignore what they did. You need to learn about cognitive science, THEN come back and argue about it. Richard Loosemore. Mike Tintner wrote: Richard, I don't think I'm not getting it at all. What you have here is a lot of good questions about how the graphics level of processing that I am proposing, might work. And I don't have the answers, and haven't really thought about them yet. What I have proposed is a general idea loosely outlining 3 levels of processing. Now if it's right, that alone is valuable. And there is at least some evidence to think it might be - starting with the strange fact that blind people produce graphics drawings, and the abundance of graphics sign systems, plus, although I didn't really deal with this, that humans do have difficulties understanding abstract verbal statements. I obviously haven't had time to demonstrate it to you,. but the idea does start to impose order on our sign systems and it's quite hard just to do that. All scientific ideas and theories only go so far and spell out things in limited detail. The fact that they are not more detailed is NOT per se an objection to them. People objected to Newton - [no, I am NOT comparing this idea in any way with his work] - because he didn't spell out how gravity worked. He didn't have to. What he showed about gravitational attraction was enough. I can't see that ANY of your questions pose an immense brick wall. If you were able to argue, for argument's sake: look, the human brain simply can't handle graphics outlines, only symbolic formulae that WOULD be a brick wall. Just consider your questions again. If I ask you, for example, to visualise a graphic of a man, and a penny you will, I suggest, do it. Your brain WILL produce relevant graphics. Now how did it do that? Why did it pick those particular graphics, given that you have vast numbers available to you? Hey, neither you nor I have the answer to how it did that (although we can think about it another time). But on one level, what does it matter? The point is: IT DID IT Your brain was not stymied, as your questions seem to imply it should be; it just went ahead. Similarly, how does the brain achieve visual object recognition? How does it manage to recognize cats and dogs? What templates does it use? How does it manage to select a particular cat template, when it may well have hundreds? I think we can be confident that it does use a template or templates one way or another. Perhaps it just grabs the nearest one at neuronal hand. (And BTW I'd be v.. interested to discuss all this in another thread). But, whatever, the brain does it. ... But if I were to be guided by the spirit of your objections, I would, say: hey I can think no more about this, the whole idea is ridiculous. Ditto re your objections as to how the brain could create moving graphics as I propose. No, I don't know exactly how it does it. Here's a frame from a dream of mine - a man with a beard on flame, in a check shirt, lying on the ground. I doubt that I have ever seen that bearded head, with that check shirt, lying in that posture, let alone on flame. The brain combined four new elements in a flash in a new moving picture. If it can do that, there is no reason as yet to think that it can't create moving graphics, or moving images if necessary to test out sentences as I propose. But re your mental models, I'm just asking, what on earth do you and others mean? I'm sure whatever you're proposing is possible, I'd just like to know what it is - and if I'm confused, i.e. find the whole concept vague, then I'm pretty confident you and everyone else are also confused - because, according to my theory, (and I believe this is true), the brain DEMANDS to have concepts like that make sense. It complains - positively aches to a greater or lesser degree - if they don't, and yours will have already. (Repeat: there is no a priori objection to the concept of mental models). There is an irony - you have just asked fifteen or so questions of my graphics idea, and not one of mental models. P.S. A personal comment here - it's offered as an intuitive response, not a reasoned judgment, if it's no use or wong, screw it. You have just offered an awful lot of what are actuially constructive suggestions and proposals for further thought, as if they were damning objections. I felt intuitively that you were dong something similar in trying to define intelligence - trying to take things to minute pieces - and in the end, turning an initially constructive drive into a negative conclusion. Like I said, a purely intuitive response, and my apologies if it's wrong or no use. There is ONE BIG THING HERE THAT YOU ARE NOT GETTING. If you were to sit down and try to implement an actual
Re: [agi] The Advantages of a Conscious Mind
My comment stemmed from my experience as a professional cognitive scientist. Please don't pull this kind of stunt. Mike Tintner wrote: Richard, Welcome to the Virtual Home for the NCSU Cognitive Science Program! Cognitive Science is an exciting area of interdisciplinary research that seeks to understand what is arguably the final mystery within the universe -- the nature and evolution of mind. Cognitive Science programs exist across the globe, typically represented by a broad range of faculty who specialize in areas like Psychology and Neuroscience, Linguistics and Psycholinguistics, Computer Science and Robotics, as well as Logic and the Philosophy of Mind. This interdisciplinary perspective is necessary, since contemporary theories of mind incorporate ideas from several disciplines. Thus the mind is usefully modeled as a rational agent, a logical system, a computer, a psycholinguistic device, and a brain whose psychological functions evolved naturally over time. Accordingly, North Carolina State University has its own Cognitive Science Program, administered by the Department of Philosophy Religion, and supported by a strong faculty drawn from the fields of Psychology, Neurobiology, Computer Science, Linguistics, and Philosophy. - Original Message - From: Richard Loosemore [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, May 06, 2007 9:09 PM Subject: Re: [agi] The Advantages of a Conscious Mind Mike Tintner wrote: Cognitive science treats humans as thinking like computers - rationally, if boundedly rationally. Which part of cognitive science treats humans as thinking irrationally, as I have described ? (There may be some misunderstandings here which hve to be ironed out, but I don't think my claim at all outrageous or less than obvious). All the social sciences treat humans as thinking rationally. It is notorious that this doesn't fit the reality - especially for example in economics. But the basic attitude is: well, it's the best model we've got. It is hard to argue with you when you make statements that so flagrantly contradict the facts: pick up a textbook of cognitive psychology (my favorite is Eysenck and Keane, but you can try John Anderson...) and you will find some chapters that specifically discuss the experimental evidence for the fact that humans do not generally think in rational ways. They study the irrationality, so how could they possibly assume that humans are rational like computers? These people would not for one minute go along with your statement that they assume that humans think like computers. That term rational is crucial. I am using it the way everyone in cognitive science uses it. Which part of cognitive science treats humans as thinking irrationally? Egads: all of it! Richard Loosemore. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Richard, I have taken your point that you are pissed off with me do not wish to talk to me. However, you are being unwarrantedly insulting to me, if you think I am pulling a stunt. I was making a genuinely meant point - it is no problem to produce an endless series of cognitive science definitions like that below, which stress that it treats the human mind as a rational agent. I did it, because I genuinely believe what I an saying - and I argue genuinely throughout, not cheaply or nastily, and from commitment. By all means disagree or think me stupid, naive, whatever. But you are not entitled to take that tone. It's OK, you don't need to reply. My comment stemmed from my experience as a professional cognitive scientist. Please don't pull this kind of stunt. Mike Tintner wrote: Richard, Welcome to the Virtual Home for the NCSU Cognitive Science Program! Cognitive Science is an exciting area of interdisciplinary research that seeks to understand what is arguably the final mystery within the universe -- the nature and evolution of mind. Cognitive Science programs exist across the globe, typically represented by a broad range of faculty who specialize in areas like Psychology and Neuroscience, Linguistics and Psycholinguistics, Computer Science and Robotics, as well as Logic and the Philosophy of Mind. This interdisciplinary perspective is necessary, since contemporary theories of mind incorporate ideas from several disciplines. Thus the mind is usefully modeled as a rational agent, a logical system, a computer, a psycholinguistic device, and a brain whose psychological functions evolved naturally over time. Accordingly, North Carolina State University has its own Cognitive Science Program, administered by the Department of Philosophy Religion, and supported by a strong faculty drawn from the fields of Psychology, Neurobiology, Computer Science, Linguistics, and Philosophy. - Original Message - From: Richard Loosemore [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, May 06, 2007 9:09 PM Subject: Re: [agi] The Advantages of a Conscious Mind Mike Tintner wrote: Cognitive science treats humans as thinking like computers - rationally, if boundedly rationally. Which part of cognitive science treats humans as thinking irrationally, as I have described ? (There may be some misunderstandings here which hve to be ironed out, but I don't think my claim at all outrageous or less than obvious). All the social sciences treat humans as thinking rationally. It is notorious that this doesn't fit the reality - especially for example in economics. But the basic attitude is: well, it's the best model we've got. It is hard to argue with you when you make statements that so flagrantly contradict the facts: pick up a textbook of cognitive psychology (my favorite is Eysenck and Keane, but you can try John Anderson...) and you will find some chapters that specifically discuss the experimental evidence for the fact that humans do not generally think in rational ways. They study the irrationality, so how could they possibly assume that humans are rational like computers? These people would not for one minute go along with your statement that they assume that humans think like computers. That term rational is crucial. I am using it the way everyone in cognitive science uses it. Which part of cognitive science treats humans as thinking irrationally? Egads: all of it! Richard Loosemore. - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?; -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.467 / Virus Database: 269.6.4/790 - Release Date: 05/05/2007 10:34 - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936
Re: [agi] The Advantages of a Conscious Mind
Pei, I assumed your system is determinisitc from your posts, not your papers. So I'm still really, genuinely confused by your position. You didn't actually answer my question (unless I've missed something in all these posts) re how your system could have a choice and yet not be arbitrary at all. Listen, you can define your system any which way you like. Why not do it simply and directly? A free system can decide at a given point, either of two or multiple ways, - in my example, to Buy, Sell or Hold. A deterministic system at that same point, will have only one option. It will have, say, to decide to Sell. Which is your system? (Philosophers may argue till the end of time about what is/ isn't compatibilist, incompatibilisit, etc etc but they won't define free and determined decisionmaking any differently). To answer your question, how do you know that the human mind is not deterministic in this sense? Just because you don't know a complex set of algorithms that can explain its behaviors? Yes, it is not impossible that there is some extremely complex set of determinisitic algorithms that explains everything. It is not impossible that we are all a simulation on a computer run by some advanced civilisation. (How do you know that we are not?) But there is NO EVIDENCE whatsoever that human behaviour does fall into deterministic patterns - no laws of scientific behaviour, despite hundreds of years of trying. No one can provide the slightest indication of what such a complex set of algorithms might be. And a nondeterministic programming explanation is basically simple. And fits the crazy evidence and much more. And - Occam's Razor - which kind of explanation should science go with? Re: Wrong. NARS often needs to work hard to decide between different goals, tasks, axioms and algorithms, and is not always successful in doing that. thanks for clarifying. But presumably once it is either successful or a failure in deciding its priorities, then its priorities are fixed? And is therefore determined, or not? Nor do I understand how or why your system could or would be deterministic and yet behave crazily like my dieting woman example for the whole of its life. By all means explain or point me to the passage in your work where you explain this. (Remember also re human, crazy behaviour that we're talking about people behaving in fundamentally self-contradictory ways - oscillating from what they consider virtuous to vicious behaviour their entire lives. I trust you will agree that this happens a great deal).. - Original Message - From: Pei Wang [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, May 06, 2007 11:45 PM Subject: Re: [agi] The Advantages of a Conscious Mind On 5/6/07, Mike Tintner [EMAIL PROTECTED] wrote: Pei, I don't think there's any confusion here. Your system as you describe it IS deterministic. Whether an observer might be confused by it is irrelevant. Equally the fact that it is determined by a complex set of algorithms applying to various tasks and domains and not by one task-specific algorithm, is also irrelevant. It's still deterministic. OK, let's use the word in this way. Then how do you know that the human mind is not deterministic in this sense? Just because you don't know a complex set of algorithms that can explain its behaviors? The point, presumably, is that your system has a clear set of priorities in deciding between different goals, tasks, axioms and algorithms Wrong. NARS often needs to work hard to decide between different goals, tasks, axioms and algorithms, and is not always successful in doing that. You confused the algorithms in a system that make it work with algorithms defined with respect to problem classes. Humans don't. Humans are still trying to work out what they really want, and what their priorities are between, for example, the different activities of their life, between work, sex, friendship, love, family etc. etc. Humans are designed to be in conflict about their fundamental goals throughout their lives. And that, I would contend, is GOOD design, and essential for their success and survival. Agree, but the same description is true for NARS, in principle. If there's any confusion, think about many women and dieting. They will be confronted by much the same decisions about whether to eat or not to eat on possibly thousands of occasions throughout their lives. And over and over, throughout their entire lives, they will - freely - decide now this way, now that. Yo-yoing on and off their diets. Your system, as I understand it, would never do that - would never act in such crazy, mixed up, contradictory ways. Your understanding about NARS is completely wrong. Can you tell me which publications of mine give you this impression? Or you simple assume that all deterministic systems must behave in this way? Pei - This list is sponsored by AGIRI: http://www.agiri.org/email
Re: [agi] What would motivate you to put work into an AGI project?
--- YKY (Yan King Yin) [EMAIL PROTECTED] wrote: On 5/6/07, Matt Mahoney [EMAIL PROTECTED] wrote: YKY, what do you mean by scruffie? Is that anyone who doesn't think FOPL should be the core of an AGI? Scruffies tend to think AGI consists of a large number of heterogeneous modules. Let's try to avoid this debate by saying we'll build as many modules as we see fit. Now FOPL is not the only KR language out there, but anyone can be reasonably familiar with it so it can serve as a foundational framework. If we are to agree on a KR scheme, it should be one that can be explained in 15 minutes. I don't think there is an elegant solution to AGI. First, people have been working on this for a long time, and if there was a simple solution we likely would have found it. Second, the complexity of AGI, prior to any training, is bounded by the complexity of DNA, which is quite high. Consider the complexity of programming a robot spider to weave webs, not by training, but by writing the algorithm yourself. Spiders are born with this knowledge. Then consider the complexity of a human brain compared to that of a spider. As for FOPL or probabilistic FOPL (for most x, p(x) is usually true, formalized with numeric probabilities), people have been down this path many times and it is a dead end. What theoretical insight do you have that would lead me to believe that your system would succeed where others have failed? I used to be pretty good at C and assembler hacking =) but we definitely should not worry about hardware at *this stage*. We should first focus on the algorithms. We need to keep in mind that the current version requires 10^15 bits of memory and 10^16 operations per second. Why would we evolve such large brains if there was a shortcut? I think we should not go FOSS just because we arn't confident of ourselves, or to try to avoid competition. We love our work and should go the extra miles to make it profitable. Those who're not interested in business matters can leave that to somebody else in the group. The problem with closed source is you have to pay your employees. Personally, I am not interested in making a lot of money. I already make enough to buy what I want. It is more important to have free time to pursue my interests. AGI, especially language, is one of my interests. But I don't want to build something aimlessly like Cyc. I would like to see an application, a goal in which progress can be measured. I currently use text compression for this purpose. Do you have a better idea? -- Matt Mahoney, [EMAIL PROTECTED] - This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415user_secret=fabd7936