Re: [singularity] Re: Revised version of Jaron Lanier's thought experiment.
On 28/02/2008, John G. Rose [EMAIL PROTECTED] wrote: Actually a better way to do it as getting even just the molecules right is a wee bit formidable - you need a really powerful computer with lots of RAM. Take some DNA and grow a body double in software. Then create an interface from the biological brain to the software brain and then gradually kill off the biological brain forcing the consciousness into the software brain. The problem with this approach naturally is that to grow the brain in RAM requires astronomical resources. But ordinary off-the-shelf matter holds so much digital memory compared to modern computers. You have to convert matter into RAM somehow. For example one cell with DNA is how many gigs? And cells cost a dime a billion. But the problem is that molecular interaction is too slow and cluncky. Agreed, it would be *enormously* difficult getting a snapshot at the molecular level and then doing a simulation from this snapshot. But as a matter of principle, it should be possible. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: [singularity] Re: Revised version of Jaron Lanier's thought experiment.
On 29/02/2008, Matt Mahoney [EMAIL PROTECTED] wrote: By equivalent computation I mean one whose behavior is indistinguishable from the brain, not an approximation. I don't believe that an exact simulation requires copying the implementation down to the neuron level, much less the molecular level. How do you explain the fact that cognition is exquisitely sensitive to changes at the molecular level? -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: [singularity] Re: Revised version of Jaron Lanier's thought experiment.
On 29/02/2008, Matt Mahoney [EMAIL PROTECTED] wrote: 4. The embodied, planned, personalized Turing test. Communication is not restricted to text. The machine is planted in the skull of your clone. Your friends and relatives have to decide who has the carbon-based brain. Level 4 should not require simulating every neuron and synapse. Without the constraints of slow, noisy neurons, we could use other algorithms. For example, low level visual processing such as edge and line detection would not need to be implemented as a 2-D array of identical filters. It could be implemented serially by scanning the retinal image with a window filter. Fine motor control would not need to be implemented by combining thousands of pulsing motor neurons to get a smooth average signal. The signal could be computed numerically. The brain has about 10^15 synapses, so a straightforward simulation at the neural level would require 10^15 bits of memory. But cognitive tests suggest humans have only about 10^9 bits of long term memory, suggesting that more compressed representation is possible. In any case, level 1 should be sufficient to argue convincingly that either consciousness can exist in machines, or that it doesn't in humans. I agree that it should be possible to simulate a brain on a computer, but I don't see how you can be so confident that you can throw away most of the details of brain structure with impunity. Tiny changes to neurons which make no difference to the anatomy or synaptic structure can have large effects on neuronal behaviour, and hence whole organism behaviour. You can't leave this sort of thing out of the model and hope that it will still match the original. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: [singularity] Re: Revised version of Jaron Lanier's thought experiment.
On 27/02/2008, John G. Rose [EMAIL PROTECTED] wrote: Well if you spend some time theorizing a model of a brain digitizer that operates within known physics constraints it's not an easy task getting just the molecular and atomic digital data. You have to sample over a period of time and space using photons and particle beams. This in itself interferes with the sample. Then say this sample is reconstructed within a theoretically capable computer, the computer will most likely have to operate in slow time to simulate the physics of all the atoms and molecules as the computer is itself constrained by the speed of light. I'm going this route because I don't think that it is possible to get an instantaneous reading of all the atoms in a brain, you have to reconstruct over time and space. THEN, this is ignoring the subatomic properties and forget about quantum data sample digitization I think it is impossible to get an exact copy. So this leaves you with a reconstructed approximation. Exactly how much of this would be you is unknown because any subatomic and quantum properties of you are - started from scratch - this includes any macroscopic and environmental properties of subatomic and quantum and superatomic molecular state and positioning effects. And if the whole atomic level model is started from scratch in the simulator it could disintegrate or diverge as it is all forced fit together. Your copy is an approximation of which it is unknown how close it is actually of you or if you could be even put together accurately enough in the simulator. There are some who think that all you need to simulate a brain (and effectively copy a person) is to fix it, slice it up, and examine it under a microscope to determine the synaptic structure. This is almost certainly way too crude: consider the huge difference to cognition made by small molecules in tiny concentrations, such as LSD, which do no more than slightly alter the conformation of certain receptor proteins on neurons by binding to them non-covalently. On the other hand, it is equally implausible to suppose that you have to get it right down to the subatomic level, since otherwise cosmic rays or changing the isotope composition of the brain would have a major effect, and they clearly don't. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: [singularity] Re: Revised version of Jaron Lanier's thought experiment.
On 28/02/2008, John G. Rose [EMAIL PROTECTED] wrote: I don't know if you can rule out subatomic and quantum. There seems to be more and more evidence pointing to an amount of activity going on there. A small amount of cosmic rays don't have obvious immediate gross effects but interaction is occurring. Exactly how much of it would need to be replicated is not known. You could be missing out on important psi elements in consciousness which are taken for granted :) Either way it would be approximation unless there was some way using theoretical physics where an exact instantaneous snapshot could occur with the snapshot existing in precisely equivalent matter at that instant. Well, maybe you can't actually rule it out until you make a copy and see how close it has to be to think the same as the original, but I strongly suspect that getting it right down to the molecular level would be enough. Even if quantum effects are important in consciousness (and I don't think there is any clear evidence that this is so), these would be generic quantum effects, reproduced by reproducing the molecular structure. Transistors function using quantum level effects, but you don't need to replace a particular transistor with a perfect copy to have an identically functioning electronic device. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: [singularity] Re: Revised version of Jaron Lanier's thought experiment.
On 26/02/2008, John G. Rose [EMAIL PROTECTED] wrote: There is an assumed simplification tendency going on that a human brain could be represented as a string of bits. It's easy to assume but I think that a more correct way to put it would be that it could be approximated. Exactly how close the approximation could theoretically get is entirely unknown. It's not entirely unknown. The maximum simulation fidelity that would be required is at the quantum level, which is still finite. But probably this would be overkill, since you remain you from moment to moment despite changes in your brain which are gross compared to the quantum level. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: [singularity] Re: Revised version of Jaron Lanier's thought experiment.
On 24/02/2008, Vladimir Nesov [EMAIL PROTECTED] wrote: Does 2+2=4 make a sound when there is no one around? Yes, but it is of no consequence since no one can hear it. However, if we believe that computation can result in consciousness, then by definition there *is* someone to hear it: itself. But it's still of no 'consequence', no? Of no consequence as far as anything at the level of the substrate of its implementation is concerned, no. In order to find such a computation hidden in noise we would have to do the computation all over again, using conventional means. But unless we require that the computation interact with us, that should make no difference to *it*. If the computation simulates an inputless virtual reality with conscious inhabitants, they should be no less conscious for the fact that we can't talk to them. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: Infinitely Unlikely Coincidences [WAS Re: [singularity] AI critique by Jaron Lanier]
On 21/02/2008, John Ku [EMAIL PROTECTED] wrote: By the way, I think this whole tangent was actually started by Richard misinterpreting Lanier's argument (though quite understandably given Lanier's vagueness and unclarity). Lanier was not imagining the amazing coincidence of a genuine computer being implemented in a rainstorm, i.e. one that is robustly implementing all the right causal laws and the strong conditionals Chalmers talks about. Rather, he was imagining the more ordinary and really not very amazing coincidence of a rainstorm bearing a certain superficial isomorphism to just a trace of the right kind of computation. He rightly notes that if functionalism were committed to such a rainstorm being conscious, it should be rejected. Only if it is incompatible with the world we observe. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: Infinitely Unlikely Coincidences [WAS Re: [singularity] AI critique by Jaron Lanier]
On 21/02/2008, John Ku [EMAIL PROTECTED] wrote: On 2/20/08, Stathis Papaioannou [EMAIL PROTECTED] wrote: On 21/02/2008, John Ku [EMAIL PROTECTED] wrote: By the way, I think this whole tangent was actually started by Richard misinterpreting Lanier's argument (though quite understandably given Lanier's vagueness and unclarity). Lanier was not imagining the amazing coincidence of a genuine computer being implemented in a rainstorm, i.e. one that is robustly implementing all the right causal laws and the strong conditionals Chalmers talks about. Rather, he was imagining the more ordinary and really not very amazing coincidence of a rainstorm bearing a certain superficial isomorphism to just a trace of the right kind of computation. He rightly notes that if functionalism were committed to such a rainstorm being conscious, it should be rejected. Only if it is incompatible with the world we observe. I think that's the wrong way to think about philosophical issues. It seems you are trying to import a scientific method to a philosophical domain where it does not belong. Functionalism is a view about how our concepts work. It is not tested by whether it is falisified by observations about the world. Or if you prefer, conceptual analysis does produce scientific hypotheses about the world, but the part of the world in question is within our own heads, something that we ourselves don't have transparent access to. If we had transparent access to the way our concepts work, the task of cognitive science and philosophy and along with it much of AI would be considerably easier. Our best way of testing these hypotheses at the moment is to see whether a proposed analysis would best explain our uses of the concept and our conceptual intuitions. Functionalism at least has the form of a scientific hypothesis, in that it asserts that a functionally equivalent analogue of my brain will have the same mental properties. Even though in practice it isn't empirically falsifiable we can examine it to make sure it is internally consistent, compatible with observed reality, and in keeping with the principle of Occam's razor. We should certainly be wary of a theory that sounds ridiculous, but unless it fails in one of these three areas it is wrong to dismiss it. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: Infinitely Unlikely Coincidences [WAS Re: [singularity] AI critique by Jaron Lanier]
On 19/02/2008, John Ku [EMAIL PROTECTED] wrote: Yes, you've shown either that, or that even some occasionally intelligent and competent philosophers sometimes take seriously ideas that really can be dismissed as obviously ridiculous -- ideas which really are unworthy of careful thought were it not for the fact that pinpointing exactly why such ridiculous ideas are wrong is so often fruitful (as in the Chalmers article). It doesn't sound so strange when you examine the distinction between the computation and the implementation of the computation. An analogy is the distinction between a circle and the implementation of a circle. It might be objected that it is ridiculous to argue that any irregular shape looked at with the right transformation matrix is an implementation of a circle. The objection is valid under a non-trivial definition of implementation. A randomly drawn perimeter around a vicious dog on a tether does not help you avoid getting bitten unless you have the relevant transformation matrix and can do the calculations in your head, which would be no better than having no implementation at all but just instructions on how to draw the circle de novo. Thus, implementation is linked to utility. Circles exist in the abstract as platonic objects, but platonic objects don't interact with the real world until they are implemented, and implemented in a particular useful or non-trivial way. Similarly, computations exist as platonic objects, such as Turing machines, but don't play any part in the real world unless they are implemented. There is an abstract machine adding two numbers together, but this no use to you when you are doing your shopping unless it is implemented in a useful and non-trivial way, such as in an electronic calculator or in your brain. Now, consider the special case of a conscious computation. If this computation is to interact with the real world it must fulfil the criteria for non-trivial implementation as discussed. A human being would be an example of this. But what if the computation creates an inputless virtual world with conscious inhabitants? Unless you are prepared to argue that the consciousness of the inhabitants is contingent on interaction with the real world there seems no reason to insist that the implementation be non-trivial or useful in the above sense. Consciousness would then be a quality of the abstract platonic object, as circularity is a quality of the abstract circle. I might add that there is nothing in this which contradicts functionalism, or for that matter geometry. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: Infinitely Unlikely Coincidences [WAS Re: [singularity] AI critique by Jaron Lanier]
On 19/02/2008, Richard Loosemore [EMAIL PROTECTED] wrote: Sorry, but I do not think your conclusion even remotely follows from the premises. But beyond that, the basic reason that this line of argument is nonsensical is that Lanier's thought experiment was rigged in such a way that a coincidence was engineered into existence. Nothing whatever can be deduced from an argument in which you set things up so that a coincidence must happen! It is just a meaningless coincidence that a computer can in theory be set up to be (a) conscious and (b) have a lower level of its architecture be isomorphic to a rainstorm. I don't see how the fact something happens by coincidence is by itself a problem. Evolution, for example, works by means of random genetic mutations some of which just happen to result in a phenotype better suited to its environment. By the way, Lanier's idea is not original. Hilary Putnam, John Searle, Tim Maudlin, Greg Egan, Hans Moravec, David Chalmers (see the paper cited by Kaj Sotola in the original thread - http://consc.net/papers/rock.html) have all considered variations on the theme. At the very least, this should indicate that the idea cannot be dismissed as just obviously ridiculous and unworthy of careful thought. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: Infinitely Unlikely Coincidences [WAS Re: [singularity] AI critique by Jaron Lanier]
On 18/02/2008, Richard Loosemore [EMAIL PROTECTED] wrote: The last statement you make, though, is not quite correct: with a jumbled up sequence of episodes during which the various machines were running the brain code, he whole would lose its coherence, because input from the world would now be randomised. If the computer was being fed input from a virtual reality simulation, that would be fine. It would sense a sudden change from real world to virtual world. The argument that is the subject of this thread wouldn't work if the brain simulation had to interact with the world at the level of the substrate it is being simulated on. However, it does work if you consider an inputless virtual environment with conscious inhabitants. Suppose you are now living in such a simulation. From your point of view, today is Monday and yesterday was Sunday. Do you have any evidence to support the belief that Sunday was was actually run yesterday in the real world, or that it was run at all? The simulation could have been started up one second ago, complete with false memories of Sunday. Sunday may not actually be run until next year, and the version of you then will have no idea that the future has already happened. But again, none of this touches upon Lanier's attempt to draw a bogus conclusion from his thought experiment. No external observer would ever be able to keep track of such a fragmented computation and as far as the rest of the universe is concerned there may as well be no computation. This makes little sense, surely. You mean that we would not be able to interact with it? Of course not: the poor thing will have been isolated from meanigful contact with the world because of the jumbled up implementation that you posit. Again, though, I see no relevant conclusion emerging from this. I cannot make any sense of your statement that as far as the rest of the universe is concerned there may as well be no computation. So we cannot communicate with it anymore that should not be surprising, given your assumptions. We can't communicate with it so it is useless as far as what we normally think of as computation goes. A rainstorm contains patterns isomorphic with an abacus adding 127 and 498 to give 625, but to extract this meaning you have to already know the question and the answer, using another computer such as your brain. However, in the case of an inputless simulation with conscious inhabitants this objection is irrelevant, since the meaning is created by observers intrinsic to the computation. Thus if there is any way a physical system could be interpreted as implementing a conscious computation, it is implementing the conscious computation, even if no-one else is around to keep track of it. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: [singularity] AI critique by Jaron Lanier
On 17/02/2008, Richard Loosemore [EMAIL PROTECTED] wrote: Lanier's rainstorm argument is spurious nonsense. That's the response of most functionalists, but an explanation as to why it is spurious nonsense is needed. And some such as Hans Moravec have actually conceded that the argument is valid: http://www.frc.ri.cmu.edu/~hpm/project.archive/general.articles/1998/SimConEx.98.html -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: [singularity] AI critique by Jaron Lanier
On 16/02/2008, Kaj Sotala [EMAIL PROTECTED] wrote: However, despite what is claimed, not every physical process can be interpreted to do any computation. To do such an interpretation, you have to do so after the fact: after all the raindrops have fallen, you can assign their positions formal roles that correspond to computation, but you can't *predict* what positions will be assigned what roles ahead of the time - after all, they are just randomly falling raindrops. You can't actually *use* the rainstorm to compute anything, like you could use a computer - you have to first do the computation yourself, then assign each state of the rainstorm a position that corresponds to the steps in your previous computation. Sure, you can't interact with the raindrop computation, but that doesn't mean it isn't conscious. Suppose a civilization built a computer implementing a virtual environment with conscious inhabitants, but no I/O. The computer is launched into space and the civilization is completely destroyed when its sun goes nova. A billion years later, the computer is found by another civilization which figures out how the power supply works and starts it up, firing the virtual inhabitants into life. As far as the second civilization is concerned, the activity in the computer could mean anything or nothing, like the patterns in a rainstorm. Just as the space of all possible rainstorms contains one that is isomorphic with any given computer implementing a particular program, so the space of all possible computers that an alien civilization might build contains one that is isomorphic with any sufficiently large rainstorm. It doesn't matter that manual for the computer represented by the rainstorm has been lost, or that the computer was never actually built: all that matters for the program to be implemented is that it rain. -- Stathis Papaioannou --- singularity Archives: http://www.listbox.com/member/archive/11983/=now RSS Feed: http://www.listbox.com/member/archive/rss/11983/ Modify Your Subscription: http://www.listbox.com/member/?member_id=4007604id_secret=96140713-a54b2b Powered by Listbox: http://www.listbox.com
Re: [singularity] Wrong focus?
On 01/02/2008, Mike Tintner [EMAIL PROTECTED] wrote: The philosophical conceit that we do not really know that there is a table (or a penis) in front of us, is just that - a fanciful conceit. It shows what happens when you rely on words and symbols as your sole medium of intellectual thought - as philosophers mainly do. In reality, you have no problem knowing and being sure of those objects and the world around you - except in exceptional circumstances. Why? Two reasons. First, all sensations/perceptions are continually being unconsciously tested for their reality - a process which I would have thought every AI/robotics person would take for granted. Hence your brain occasionally thinks: was that really so-and-so I saw?...or: where exactly in my foot *is* that pain? Your unconscious brain has had problems checking some perception. Secondly, your brain works by *common sense* perception and testing. We are continually testing our perceptions with all our senses and our whole body. You don't just look at things, you reach out and touch them, smell them, taste them, and confirm over and over that your perceptions are valid. (Also it's worth pointing out that since you are continually moving in relation to objects, your many different-angle shots of them are continually tested against each other for consistency). Like a good journalist, you check more than one source. Your perceptions are continually tested in a deeply embodied way - and in general v. much in touch with reality. I'm not suggesting that there is any reason to believe there is no real world out there. What I am saying is that *if* the world you perceive were due to computer-generated data at an arbitrarily high level of resolution fed into your brain, it would respond in the same way as if it were in an intact body interacting with a real environment and you would have no way of knowing what was going on. Thus your claim that it is *impossible* for an intelligence to function in a virtual environment is false. (The weaker claim that it might be easier for an intelligence to develop and function in a real environment using a robot body, for example because this is computationally cheaper than building a virtual environment of comparable richness, may yet have merit.) The other point I was trying to make is that even if the world is real, the picture of the world your brain creates from sensory data is an abstraction that exists only in the computational space that is your mind. The map is not the territory. -- Stathis Papaioannou - 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=4007604id_secret=92524190-74820c
Re: [singularity] Wrong focus?
On 31/01/2008, Mike Tintner [EMAIL PROTECTED] wrote: I think the question should reverse - I and every (most?) creature can distinguish between a real and a virtual environment. How on earth can a virtual creature make the same distinction? How can it have a body, or a continuous sense of a body? How can it have a continous map of the world, with a continuous physical sense of up/down, forward/back, heaviness/lightness? And a fairly continuous sense of time passing? How can it have a self? How can it have continuous (conflicting) emotions coursing through its body? How can it also have a continuous sense of its energy and muscles - of zest/apathy, strength/weakness, awakeness/tiredness? How can it have a sense of its posture, and muscles tight or loose? The fact is, you are already living in a virtual environment. Your brain creates a picture of the world based on sensory data. You can't *really* know what a table is, or even that there is a table there in front of you at all. All you can know is that you have particular table-like experiences, which seem to be consistently generated by what you come to think of as the external object table. There is no way to be certain that the picture in your head - including the picture you have of your own body - is generated by a real external environment rather than by a computer sending appropriately high resolution signals to fool your brain: http://en.wikipedia.org/wiki/Brain_in_a_vat -- Stathis Papaioannou - 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=4007604id_secret=92100746-21f656
Re: [singularity] Wrong focus?
On 29/01/2008, [EMAIL PROTECTED] [EMAIL PROTECTED] wrote: On Jan 28, 2008 4:36 AM, Stathis Papaioannou [EMAIL PROTECTED] wrote: Are you simply arguing that an embodied AI that can interact with the real world will find it easier to learn and develop, or are you arguing that there is a fundamental reason why an AI can't develop in a purely virtual environment? I think the answer to the above is obvious, but the more interesting question is whether it even makes sense to speak of a mind independent of some environment of interaction, whether physical or virtual. Could that just mean in the limiting case that one part of a physical object is a mind with respect to another part? -- Stathis Papaioannou - 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=4007604id_secret=90839892-81029c
Re: [singularity] Wrong focus?
On 29/01/2008, Mike Tintner [EMAIL PROTECTED] wrote: The latter. I'm arguing that a disembodied AGI has as much chance of getting to know, understand and be intelligent about the world as Tommy - a deaf, dumb and blind and generally sense-less kid, that's totally autistic, can't play any physical game let alone a mean pin ball, and has a seriously impaired sense of self , (what's the name for that condition?) - and all that is even if the AGI *has* sensors. Think of a disembodied AGI as very severely mentally and physically disabled from birth - you wouldn't do that to a child, why do it to a computer? It might be able to spout an encyclopaedia, show you a zillion photographs, and calculate a storm but it wouldn't understand, or be able to imagine/ reimagine, anything. How can you tell the difference between sensory input from a real environment and that from a virtual environment? -- Stathis Papaioannou - 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=4007604id_secret=90867376-c56f6a
Re: [singularity] I feel your pain
On 06/11/2007, Don Detrich [EMAIL PROTECTED] wrote: Will an AGI with no bio-heritage be able to feel our pain, have empathy? If not, will that make it less conscious and more dangerous? Empathy is just another function of the brain, like visual perception. Neurons involved in empathic feeling do not contain special non-computable components absent in the neurons of the visual cortex. -- Stathis Papaioannou - 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=4007604id_secret=61522846-d61d2a
Re: [singularity] CONJECTURE OR TRUTH
On 26/10/2007, albert medina [EMAIL PROTECTED] wrote: Dear Sir or Madam, The human brain knows that it does not know. This very fact prompts it to act to know, producing the innovation which has brought us to this moment in time. This is because it derives all of its vital energy from a source beyond itself. . .above it and infinitely more subtle than it. It cannot tap into that source by its own efforts because the source is non-material and is an energy which cannot be measured by any means. Ironically, the source only supplies raw energy. . .the brain may do with it what it likes (free will, constructive or destructive). The brain is not alive, nor is it conscious. It borrows all of its vital energy from the source mentioned above. That source is the reservoir of consciousness, beyond physics (of any type) and beyond metaphysics. Again, it cannot be measured by any material instrument. Man cannot produce this original, vital energy. Consciousness is not of man. . .it is used by man (and woman) and every lifeform known. It is immortal, ubiquitous and unknown. The exoteric must confront the esoteric, but it will always be defeated because of former is an effect and the latter is the cause. Sincerely, Albert You do realise that most readers of this list will regard what you have just written as nonsense not even worth rebutting? I don't intend this to be denigrating, just a description of your audience. -- Stathis Papaioannou - 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=4007604id_secret=57767605-3be7fe
Re: [singularity] John Searle...
On 26/10/2007, Allen Majorovic [EMAIL PROTECTED] wrote: It seems to me that Mr. Searles is suggesting that because some people (intelligences) are cooks, i.e. work from a set of rules they don't understand, this somehow proves that chemists, i.e. people who *do* understand the set of rules, don't, or can't, exist. If the guy with the book of rules in his lap doesn't have to understand Chinese to do the translations, does the guy who wrote the book of rules have to under Chinese in order to write it? Searle would probably say that the person who sets up the Chinese Room has to understand Chinese, but the person in the room does not. This is true, but as has been pointed out previously, it is possible for the system to understand Chinese while the individual components of the system do not. Individual neurons in a Chinese speaker's brain understand even less of the process they participate in than the person in the Chinese Room does, yet the brain as a whole understands Chinese. -- Stathis Papaioannou - 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=4007604id_secret=57829247-7a6e38
Re: [singularity] Re: CEV
On 26/10/2007, Richard Loosemore [EMAIL PROTECTED] wrote: If you build an AGI, and it sets out to discover the convergent desires (the CEV) of all humanity, it will be doing this because it has the goal of using this CEV as the basis for the friendly motivations that will henceforth guide it. But WHY would it be collecting the CEV of humanity in the first phase of the operation? What would motivate it to do such a thing? What exactly is it in the AGI's design that makes it feel compelled to be friendly enough toward humanity that it would set out to assess the CEV of humanity? The answer is: its initial feelings of friendliness toward humanity would have to be the motivation that drove it to find out the CEV. The goal state of its motivation system is assumed in the initial state of its motivation system. Hence: circular. You don't have to assume that the AI will figure out the CEV of humanity because it's friendly; you can just say that its goal is to figure out the CEV because that is what it has been designed to do, and that it has been designed to do this because its designers have decided that this a good way to ensure behaviour which will be construed as friendly. I don't see that the CEV goal would be much different to creating an AI that simply has the goal of obeying its human masters. Some of the instructions it will be given will be along the lines of AI, if you think I'll be really, really upset down the track if you carry out my instructions (where you determine what I mean by 'really, really upset' using your superior intellect), then don't carry out my instructions. If there are many AI's with many human masters, averaging out their behaviour will result in an approximation of the CEV of humanity. -- Stathis Papaioannou - 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=4007604id_secret=57845052-2cface
Re: [singularity] Towards the Singularity
On 13/09/2007, Matt Mahoney [EMAIL PROTECTED] wrote: I think the usual explanation is that the split doubles the number of universes and the number of copies of a brain. It wouldn't make any difference if tomorrow we discovered a method of communicating with the parallel universes: you would see the other copies of you who have or haven't observed the atom decay but subjectively you still have a 50% chance of finding yourself in one or other situation if you can only have the experiences of one entity at a time. If this is true, then it undermines an argument for uploading. Some assume that if you destructively upload, then you have a 100% chance of being the copy. But what if the original is killed not immediately, but one second later? In that case, you have a 50% chance of ending up the original and 50% chance of ending up the copy. If you end up the original, you then have a 100% chance of dying, which I think of as the inability to anticipate any future experiences. My preferred way of looking at these questions is to acknowledge that there is no self persisting through time in any absolute sense, but rather a set of observer moments which are only contingently related. One instance of me considers certain other instances past selves and certain other instances future selves. The future selves' experiences are anticipated while the past selves' experiences are not even though both have equal claim to being me. Worse, the future selves' experiences are anticipated even if they occur in the actual present or past, as in a block universe or in a simulation running backwards. If we attempt to impose the naturally evolved sense of self onto unnatural scenarios involving uploading and duplication, the result is what I have been trying to describe in terms of survival and subjective probabilities. I don't actually believe that my self somehow transfers into my upload; I know that as a matter of fact, I will die. But neither do I believe that in everyday life my self transfers into the next instantiation of my brain with every passing moment. I am willing to admit that I live only transiently and the sense of a persisting self is a kind of illusion. However, I would like that illusion to continue in the same way as long as possible, and destructive uploading will do just that. These problems go away if you don't assume consciousness exists. Then the question is, if I encounter someone that claims to be you, what is the probability that I encountered your copy? I can ask the same question for myself: if I find myself thinking I am me, what is the probability that I am the copy? -- Stathis Papaioannou - 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=4007604id_secret=41463669-259d6d
Re: [singularity] Towards the Singularity
On 10/09/07, Matt Mahoney [EMAIL PROTECTED] wrote: No, it is not necessary to destroy the original. If you do destroy the original you have a 100% chance of ending up as the copy, while if you don't you have a 50% chance of ending up as the copy. It's like probability if the MWI of QM is correct. No, you are thinking in the present, where there can be only one copy of a brain. When technology for uploading exists, you have a 100% chance of becoming the original and a 100% chance of becoming the copy. It's the same in no collapse interpretations of quantum mechanics. There is a 100% chance that a copy of you will see the atom decay and a 100% chance that a copy of you will not see the atom decay. However, experiment shows that there is only a 50% chance of seeing the atom decay, because the multiple copies of you don't share their experiences. The MWI gives the same probabilistic results as the CI for any observer. So if your brain is a Turing machine in language L1 and the program is recompiled to run in language L2, then the consciousness transfers? But if the two machines implement the same function but the process of writing the second program is not specified, then the consciousness does not transfer because it is undecidable in general to determine if two programs are equivalent? It depends on what you mean by implements the same function. A black box that emulates the behaviour of a neuron and can be used to replace neurons one by one, as per Hans Moravec, will result in no alteration to consciousness (as shown in David Chalmers' fading qualia paper: http://consc.net/papers/qualia.html), so total replacement by these black boxes will result in no change to consciousness. It doesn't matter what is inside the black box, as long as it is functionally equivalent to the biological tissue. On the other hand... I mean implements the same function in that identical inputs result in identical outputs. I don't insist on a 1-1 mapping of machine states as Chalmers does. I doubt it makes a difference, though. Chalmers' argument works for identical outputs for identical inputs. Also, Chalmers argues that a machine copy of your brain must be conscious. But he has the same instinct to believe in consciousness as everyone else. My claim is broader: that either a machine can be conscious or that consciousness does not exist. I think Chalmers' claim is that either a machine can be conscious or else some sort of weird substance dualism is the case. I'm not sure I understand what you mean when you say consciousness does not exist. Even if it's just an epiphenomenon, nothing but what it feels like to process certain kinds of information, there is a sense in which it exists. Otherwise it's like saying multiplication doesn't exist because it's just repeated addition. -- Stathis Papaioannou - 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=4007604id_secret=40100384-1dbeb8
Re: [singularity] Uploaded p-zombies
On 10/09/07, Vladimir Nesov [EMAIL PROTECTED] wrote: As a final paradoxical example, if implementation Z is nothing, that is it comprises no matter and information ar all, there still is a correspondence function F(Z)=S which supposedly asserts that Z is X's upload. There can even be a feature extractor (which will have to implement functional simulation of S) that works on an empty Z. What is the difference from subjective experience simulation point of view between this empty Z and a proper upload implementation? A profound point that anyone who believes in computationalism has to address. The only way I can think of to keep computationalism and remain consistent is to drop the thesis that consciousness supervenes on physical activity. Rather, we can say that consciousness is a Platonic object that supervenes on an abstract machine, with physical activity such as that of brains or computers being simply a realization of an abstract machine, not actually contributing or detracting from the measure of a particular consciousness, since you can't change the measure of an abstract mathematical object by having more or fewer physical examples of it. This would leave no place for a concrete physical world: everything we see is a subset of all possible simulations running on an abstract machine. Certainly, this is weird, but the alternative would seem to be that the mind is not Turing emulable. -- Stathis Papaioannou - 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=4007604id_secret=40103766-61ed4d
Re: [singularity] Towards the Singularity
On 11/09/2007, Matt Mahoney [EMAIL PROTECTED] wrote: No, you are thinking in the present, where there can be only one copy of a brain. When technology for uploading exists, you have a 100% chance of becoming the original and a 100% chance of becoming the copy. It's the same in no collapse interpretations of quantum mechanics. There is a 100% chance that a copy of you will see the atom decay and a 100% chance that a copy of you will not see the atom decay. However, experiment shows that there is only a 50% chance of seeing the atom decay, because the multiple copies of you don't share their experiences. The MWI gives the same probabilistic results as the CI for any observer. The analogy to the multi-universe view of quantum mechanics is not valid. In the multi-universe view, there are two parallel universes both before and after the split, and they do not communicate at any time. When you copy a brain, there is one copy before and two afterwards. Those two brains can then communicate with each other. I think the usual explanation is that the split doubles the number of universes and the number of copies of a brain. It wouldn't make any difference if tomorrow we discovered a method of communicating with the parallel universes: you would see the other copies of you who have or haven't observed the atom decay but subjectively you still have a 50% chance of finding yourself in one or other situation if you can only have the experiences of one entity at a time. The multi-universe view cannot be tested. The evidence in its favor is Occam's Razor (or its formal equivalent, AIXI, assuming the universe is a computation). The important point for this argument is just that the multiverse idea cannot be tested. Whether there is one or many universes in which all outcomes occur, the probabilities work out the same. The view that you express is that when a brain is copied, one copy becomes human with subjective experience and the other becomes a p-zombie, but we don't know which one. The evidence in favor of this view is: That's not what I meant at all, if I gave that impression. Both copies are conscious and both copies have equal claim to being a continuation of the original, but each copy can only experience being one person at a time. Given this, the effect is of ending up one or other copy with equal probability, the same as if only one or other copy were created. - Human belief in consciousness and subjective experience is universal and accepted without question. Any belief programmed into the brain through natural selection must be true in any logical system that the human mind can comprehend. - Out of 6 billion humans, no two have the same memory. Therefore by induction, it is impossible to copy consciousness. (I hope that you can see the flaws in this evidence). This view also cannot be tested, because there is no test to distinguish a conscious human from a p-zombie. Unlike the multi-universe view where a different copy becomes conscious in each universe, the two universes would continue to remain identical. I think it's unlikely that p-zombies are physically possible (although they are logically possible). I don't see any problem with having multiple copies of a given consciousness. I don't see any problem with testing for consciousness, since we all perform the test on ourselves every waking moment; it's just that there are technical difficulties performing a direct test on someone else. -- Stathis Papaioannou - 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=4007604id_secret=40425827-752d41
Re: [singularity] Uploaded p-zombies
On 10/09/07, Matt Mahoney [EMAIL PROTECTED] wrote: --- Vladimir Nesov [EMAIL PROTECTED] wrote: I intentionally don't want to exactly define what S is as it describes vaguely-defined 'subjective experience generator'. I instead leave it at description level. If you can't define what subjective experience is, then how do you know it exists? If it does exist, then is it a property of the computation, or does it depend on the physical implementation of the computer? How do you test for it? You don't need to define it to know that it exists and to be able to test for it. I know what red looks like, I can test if something is red by looking at it, and scientific instruments can be used to determine the range of wavelengths that would qualify as red in my perception (Is that red? Yes OK, I'll write down 650nm). This defines criteria for producing the experience red, but it does not define or describe the experience red such that a blind person person would know what I was talking about. More generally, we can discuss in detail what it would take to produce consciousness (brains, transistors, environment etc.) leaving consciousness as something only implicitly understood by those who have it. -- Stathis Papaioannou - 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=4007604id_secret=40026339-7367f9
Re: [singularity] Towards the Singularity
On 09/09/07, Matt Mahoney [EMAIL PROTECTED] wrote: Your dilemma: after you upload, does the original human them become a p-zombie, or are there two copies of your consciousness? Is it necessary to kill the human body for your consciousness to transfer? I have the same problem in ordinary life, since the matter in my brain from a year ago has almost all dispersed into the biosphere. Even the configuration matter in my current brain, and the information it represents, only approximates that of my erstwhile self. It's just convenient that my past selves naturally disintegrate, so that I don't encounter them and fight it out to see which is the real me. We've all been through the equivalent of destructive uploading. What if the copy is not exact, but close enough to fool others who know you? Maybe you won't have a choice. Suppose you die before we have developed the technology to scan neurons, so family members customize an AGI in your likeness based on all of your writing, photos, and interviews with people that knew you. All it takes is 10^9 bits of information about you to pass a Turing test. As we move into the age of surveillance, this will get easier to do. I bet Yahoo knows an awful lot about me from the thousands of emails I have sent through their servers. There is no guarantee that something which behaves the same way as the original also has the same consciousness. However, there are good arguments in support of the thesis that something which behaves the same way as the original as a result of identical or isomorphic brain structure also has the same consciousness as the original. (Same in this context does not mean one and the same, any more than I am one and the same as my past selves.) -- Stathis Papaioannou - 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=4007604id_secret=39896278-cab09e
Re: [singularity] Towards the Singularity
On 09/09/07, Matt Mahoney [EMAIL PROTECTED] wrote: Your dilemma: after you upload, does the original human them become a p-zombie, or are there two copies of your consciousness? Is it necessary to kill the human body for your consciousness to transfer? I have the same problem in ordinary life, since the matter in my brain from a year ago has almost all dispersed into the biosphere. Even the configuration [of] matter in my current brain, and the information it represents, only approximates that of my erstwhile self. It's just convenient that my past selves naturally disintegrate, so that I don't encounter them and fight it out to see which is the real me. We've all been through the equivalent of destructive uploading. So your answer is yes? No, it is not necessary to destroy the original. If you do destroy the original you have a 100% chance of ending up as the copy, while if you don't you have a 50% chance of ending up as the copy. It's like probability if the MWI of QM is correct. There is no guarantee that something which behaves the same way as the original also has the same consciousness. However, there are good arguments in support of the thesis that something which behaves the same way as the original as a result of identical or isomorphic brain structure also has the same consciousness as the original. So if your brain is a Turing machine in language L1 and the program is recompiled to run in language L2, then the consciousness transfers? But if the two machines implement the same function but the process of writing the second program is not specified, then the consciousness does not transfer because it is undecidable in general to determine if two programs are equivalent? It depends on what you mean by implements the same function. A black box that emulates the behaviour of a neuron and can be used to replace neurons one by one, as per Hans Moravec, will result in no alteration to consciousness (as shown in David Chalmers' fading qualia paper: http://consc.net/papers/qualia.html), so total replacement by these black boxes will result in no change to consciousness. It doesn't matter what is inside the black box, as long as it is functionally equivalent to the biological tissue. On the other hand... On the other hand, your sloppily constructed customized AGI will insist that it is a conscious continuation of your life, even if 90% of its memories are missing or wrong. As long as the original is dead then nobody else will notice the difference, and others seeing your example will have happily discovered the path to immortality. That could be like an actor taking my place. Admittedly it might be difficult to tell us apart, but that is no guarantee of survival. Arguments based on the assumption that consciousness exists always lead to absurdities. But belief in consciousness is instinctive and universal. It cannot be helped. The best I can do is accept both points of view, realize they are inconsistent, and leave it at that. What is the difference between really being conscious and only thinking that I am conscious? -- Stathis Papaioannou - 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=4007604id_secret=39906466-1ec335
Re: [singularity] Towards the Singularity
On 08/09/07, Matt Mahoney [EMAIL PROTECTED] wrote: I agree this is a great risk. The motivation to upload is driven by fear of death and our incorrect but biologically programmed belief in consciousness. The result will be the extinction of human life and its replacement with godlike intelligence, possibly this century. The best we can do is view this as a good thing, because the alternative -- a rational approach to our own intelligence -- would result in extinction with no replacement. If my upload is deluded about its consciousness in exactly the same way you claim I am deluded about my consciousness, that's good enough for me. -- Stathis Papaioannou - 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=4007604id_secret=39793371-22c02b
Re: [singularity] Re: A consciousness non-contradiction
On 20/08/07, Aleksei Riikonen [EMAIL PROTECTED] wrote: On 8/20/07, Matt Mahoney [EMAIL PROTECTED] wrote: --- Samantha Atkins [EMAIL PROTECTED] wrote: Huh? Are you conscious? I believe that I am, in the sense that I am not a p-zombie. http://en.wikipedia.org/wiki/Philosophical_zombie I also believe that the human brain can be simulated by a computer, which has no need for a consciousness in this sense. I realize these beliefs are contradictory, but I just leave it at that. They are not contradictory, until it is demonstrated that a perfect simulation/copy of a human brain *isn't* conscious. For the time being, it is certainly rational to expect such a copy to be conscious, since we the original copies are conscious. It does seem that consciousness is not necessary to produce an equally capable information processing mechanism as the human brain, but through introspection it obvious that these particular information processing mechanisms that we are are indeed conscious, and hence it is rational to expect a perfect enough copy to be conscious too. Suppose a part of your brain were replaced with a cyborg implant that exactly emulated the behaviour of the missing neural tissue: accepted inputs from the surrounding neurons, computed all the biochemical reactions that would occur had the implant not been in place, and sent outputs to the surrounding neurons. This would have to be possible, even if it presented insurmountable practical difficulties, given that brain chemistry is computable, and there is no reason to think that it isn't. Say this implant involves a large part of your visual cortex. Someone holds up their hand and asks, How many fingers? Without the implant, you would have said, Three. With the implant, therefore, you say, three: same external behaviour because the implant perfectly simulates the missing brain tissue, by our original assumption. Now, suppose that the implant *isn't conscious but only behaves as if it's conscious*. In other words, you now have a zombie visual cortex which sends impulses to your motor cortex making you say you see three fingers when in fact you are thinking, Oh my God I've gone blind!. What's worse, you can't scream or shake your head or even increase your heart rate because (remember) your zombie implant perfectly simulates the external behaviour of the original brain, and screaming and shaking your head and increasing your heart rate are certainly external behaviours. The conclusion would then have to be that either replacing enough neurons for you to notice that they are missing would cause a bizarre and nightmarish decoupling between consciousness and external behaviour, or else a cyborg replacement that was functionally equivalent to the original brain would also have to result in equivalent consciousness. This is an account of David Chalmer's fading qualia argument in favour of computationalism. -- Stathis Papaioannou - 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=4007604id_secret=33563088-082508
[singularity] Is the world a friendly or unfriendly AI?
Despite the fact that it seems to lack a single unified consciousness the world of humans and their devices behaves as if it is both vastly more intelligent and vastly more powerful than any unassisted individual human. If you could build a machine that ran a planet all by itself just as well as 6.7 billion people can, doing all the things that people do as fast as people do them, then that would have to qualify as a superintelligent AI even if you can envisage that with a little tweaking it could be truly godlike. The same considerations apply to me in relation to the world as apply to an ant relative to a human or to humanity relative to a vastly greater AI (vastly greater than humanity, not just vastly greater than a human). If the world decided to crush me there is nothing I could do about it, no matter how strong or fast or smart I am. As it happens, the world is mostly indifferent to me and some parts of it will destroy me instantly if I get in their way: if I walk into traffic only a few metres from where I am sitting. But even if it wanted to help me there could be problems: if the world decided it wanted to cater to my every command I might request paperclips and it might set about turning everything into paperclip factories, or if it wanted to make me happy it might forcibly implant electrodes in my brain. And yet, I feel quite safe living with this very powerful, very intelligent, potentially very dangerous entity all around me. Should I worry more as the world's population and technological capabilities increase further, rendering me even weaker and more insignificant in comparison? -- Stathis Papaioannou - 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=4007604id_secret=22112602-8dbf63
Re: [singularity] ESSAY: Why care about artificial intelligence?
On 13/07/07, Mike Tintner [EMAIL PROTECTED] wrote: Comment: You and others seem to be missing the point, which obviously needs spelling out. There is no way of endowing any agent with conceptual goals that cannot be interpreted in ways opposite to the designer's intentions - that is in the general, abstract nature of language symbolic systems. For example, the general, abstract goal of helping humanity can legitimately in particular, concrete situations be interpreted as wiping out the entire human race (bar, say, two) - for the sake of future generations. With humans we have always had to deal with not only honest misunderstanding but also frank treachery or malice. I would hope that treachery is less likely to be a problem with AI's, but surely the risk that an AI will do something bad as a result of the treachery of another human will be at least as great as the risk that it will do something bad due to unforeseen consequences of following instructions. Our defence against such a threat will then be the same as our general defence against threats from other humans: that no single agent will rapidly be able to rise to a level of power so as to be able to dominate all of the others. -- Stathis Papaioannou - 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=4007604id_secret=20956299-e12a67
Re: [singularity] critiques of Eliezer's views on AI
On 05/07/07, Heartland [EMAIL PROTECTED] wrote: At this point it might be useful to think about why we lack access to subjective experience of a different person. (Yes, I'm assuming my neighbor is a different person. If you don't agree with this assumption (and if you don't please tell me why), this will not work.) There is an overwhelming temptation that so many succumb to to think that lack of access to subjective experience of another person is due to differences between types of two brain structures (patterns). In reality, it's *all* due to the fact that any two minds (regardless of whether they share the same type or not) are two instances of a physical process. Your life does not end when your neighbor dies and vice versa. This is understandable, verifiable and obvious. What's missing is the realization that your instance of subjective experience (process) is as isolated from your copy's instance as it is isolated from your neighbor's instance. Agreed so far. This is why I don't expect *this* life to continue through a different instance even though the next instance might occur on the same mindware a minute after the previous instance expires. But different moments of existence in a single person's life can also be regarded as different instances. This is strikingly obvious in block universe theories of time, which are empirically indistinguishable from linear time models precisely because our conscious experience would seem continuous in either case. The same is the case in the MWI of QM: multiple instances of you are generated every moment, and while before they are generated you consider that you could equally well become any of these instances, after they are generated all but one of the instances become other. This time asymmetry of self/other when it comes to copies is mirrored in duplication thought experiments, where before the duplication you can anticipate the experiences of either copy but post-duplication the copies will fight it out among themselves even though they are identical. As you suggest, it is only possible to be one person at a time. However, when your copy lies in your subjective future you expect to become him, or at random one of the hims if there is more than one. If you could travel through time, or across parallel universes, you would come across just the sort of conflict between copies that you describe, because you can only be one instance of a person in time and space. There's no such thing as pause in execution of a single instance of process. There can only be one instance before the pause and another one after the pause. Create and destroy are only operations on instances of processes. You have just arbitrarily decided that to define an instance in this way. That's OK, it's your definition, but most people would say that it therefore means a single person can exist across different instances. Moreover, it is considered possible that time is discrete, so that the universe pauses after each planck interval and nothing happens between the pauses. Would this mean that you only survive for a planck interval? I also believe you cannot consistently maintain that life continues through replacement atoms in the usual physiological manner but would not continue if a copy were made a different way. Why should it make a difference if 1% of the atoms are replaced per year or 99% per second, if the result in each case is atoms in the correct configuration? If 99% replacement is acceptable why not 100% instantaneous replacement? If 100% *instantaneous* replacement doesn't interrupt the process then we're dealing with the same instance of life and I see no problem with that. Also, as I had pointed out to you few times before, any process is necessarily defined across time interval 0 so counterarguments based on cases where time interval = 0 are not valid. In other words, it takes some time to kill the process. In real life, atoms are replaced on the fly and this takes some time. So I will rephrase the question: does it make a difference if 1%, 99% or 100% of the atoms in a person are simultaneously replaced at the same rate as single atom replacement occurs in real life? Given that these replacements can be expected to result in partial or complete (temporary) disruption of physiological processes, what percentage of replacement results in a new instance being created? What about the case where one hemisphere of the brain is replaced while the other is left alone, giving two instances communicating through the corpus callosum: do you predict that they will consider themselves two different people or will there just be one person who thinks that nothing unusual has happened, even though he is now a hybrid? -- Stathis Papaioannou - 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=4007604id_secret=12026607-191d22
Re: [singularity] critiques of Eliezer's views on AI
On 04/07/07, Heartland [EMAIL PROTECTED] wrote: Right, but Heartland disagrees, and the post was aimed at him and others who believe that a copy isn't really you. Stathis, I don't subscribe to your assertion that a person after gradual replacement of atoms in his brain is a copy. Yes, I'm aware of that, and my question was, if I were to assert that after gradual replacement of a certain proportion of atoms a person is no longer the same person, what counterargument would you use? You can't argue that it's false because you feel yourself to be the same person despite atom replacement, since that argument also applies in the case of process interruption. -- Stathis Papaioannou - 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=4007604id_secret=10835325-8a30fd
Re: [singularity] critiques of Eliezer's views on AI
On 30/06/07, Heartland [EMAIL PROTECTED] wrote: Objective observers care only about the type of a person and whether it's intantiated, not about the fate of its instances (because, frankly, they're not aware of the difference between the type and an instance). But since I know better, I would be sad about dead instances. The point is whether I'm sad/upset or not about a fact not does change that fact. Most people would be upset by the prospect of their death, and if death is interruption of brain processes, they should be upset by this. However, it is your definition of death which is at issue. If someone chose to objectively define death as replacement of a certain proportion of the matter in a person's brain, what argument would you use against this definition? -- Stathis Papaioannou - 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=4007604id_secret=9939388-6c0916
Re: [singularity] critiques of Eliezer's views on AI
On 04/07/07, Tom McCabe [EMAIL PROTECTED] wrote: Using that definition, everyone would die at an age of a few months, because the brain's matter is regularly replaced by new organic chemicals. I know that, which is why I asked the question. It's easy enough to give a precise and objective definition of death but completely miss the point of the meaning of death. -- Stathis Papaioannou - 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=4007604id_secret=10742300-135f57
Re: [singularity] AI concerns
On 02/07/07, Jef Allbright [EMAIL PROTECTED] wrote: While I agree with you in regard to decoupling intelligence and any particular goals, this doesn't mean goals can be random or arbitrary. To the extent that striving toward goals (more realistically: promotion of values) is supportable by intelligence, the values-model must be coherent. I'm not sure what you mean by coherent. If I make it my life's work to collect seashells, because I want to have the world's biggest seashell collection, how does that rate as a goal in terms of arbitrariness and coherence? -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] AI concerns
On 02/07/07, Tom McCabe [EMAIL PROTECTED] wrote: It would be vastly easier for a properly programmed AGI to decipher what we meant that it would be for humans. The question is- why would the AGI want to decipher what human mean, as opposed to the other 2^1,000,000,000 things it could be doing? It would be vastly easier for me to build a cheesecake than it would be for a chimp, however, this does not mean I spend my day running a cheesecake factory. Realize that, for a random AGI, deciphering what humans mean is not a different kind of problem than factoring a large number. Why even bother? If it's possible to design an AI that can think at all and maintain coherent goals over time, then why would you design it to choose random goals? Surely the sensible thing is to design it to do what I say and what I mean, to inform me of the consequences of its actions as far as it can predict them, to be truthful, and so on. Maybe it would still kill us all through some oversight (on our part and on the part of the large numbers of other AI's all trying to do the same thing, and keep an eye on each other), but then if a small number of key people go psychotic simultaneously, they could also kill us all with nuclear weapons. There are no absolute guarantees, but I don't see why an AI with power should act more erratically than a human with power. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] AI concerns
On 01/07/07, Eugen Leitl [EMAIL PROTECTED] wrote: that disabling your opponent would be helpful, it's because the problem it is applying its intelligence to is winning according to the formal rules of chess. Winning at any cost might look like the same problem to us vague humans, but it isn't. It doesn't matter how you win the game, but that you win the game. Anyone who doesn't understand that is not vague, he's dead, long-term. But the constraints of the problem are no less a legitimate part of the problem than the rest of it. If you're free to solve the problem win at chess using just the formal rules of the game by redefining it to win at chess using any means possible, you may as well redefine it to go sit on the beach and read a book. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] AI concerns
On 02/07/07, Eugen Leitl [EMAIL PROTECTED] wrote: But in the final analysis, the AI would be able to be implemented as code in a general purpose language on a general purpose computer with Absolutely not. Possibly, something like a silicon compiler with billions to trillions asynchronous systems. Certainly not your grandfather's computer. sufficient storage. Any lack in efficiency of such an approach would eventually be overcome by brute force increase in processing speed. No, there are physical limits. You have to go asynchronous OOP, and fine-grained sea of gates. Even current approaches are 3d torus of nodes of microkernel OS, soon with FPGAs Co. But in the end, the AI will be Turing emulable, which means you can run it on a general purpose computer with sufficient memory. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] AI concerns
On 02/07/07, Tom McCabe [EMAIL PROTECTED] wrote: The AGI doesn't care what any human, human committee, or human government thinks; it simply follows its own internal rules. Sure, but its internal rules and goals might be specified in such a way as to make it refrain from acting in a particular way. For example, if it has as its most important goal obeying the commands of humans, that's what it will do. It won't try to find some way out of it, because that assumes it has some other goal which trumps obeying humans. If it is forced to randomly change its goals at regular intervals then it might become disobedient, but not otherwise. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] AI concerns
On 01/07/07, Tom McCabe [EMAIL PROTECTED] wrote: An excellent analogy to a superintelligent AGI is a really good chess-playing computer program. The computer program doesn't realize you're there, it doesn't know you're human, it doesn't even know what the heck a human is, and it would gladly pump you full of gamma radiation if it made you a worse player. Nevertheless, it is still intelligent, more so than you are: it can foresee everything you try and do, and can invent new strategies and use them to come out of nowhere and beat you by surprise. Trying to deprive a superintelligent AI of free will is as absurd as Gary Kasparov trying to deny Deep Blue free will within the context of the gameboard. But Deep Blue wouldn't try to poison Kasparov in order to win the game. This isn't because it isn't intelligent enough to figure out that disabling your opponent would be helpful, it's because the problem it is applying its intelligence to is winning according to the formal rules of chess. Winning at any cost might look like the same problem to us vague humans, but it isn't. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] AI concerns
On 01/07/07, Tom McCabe [EMAIL PROTECTED] wrote: But Deep Blue wouldn't try to poison Kasparov in order to win the game. This isn't because it isn't intelligent enough to figure out that disabling your opponent would be helpful, it's because the problem it is applying its intelligence to is winning according to the formal rules of chess. Exactly. The formal rules of chess say stuff about where to put pawns and knights; they're analogous to the laws of physics. They don't say anything about poisoning the opposing player. If you try to build in a rule about poisoning the player, the chess program will shoot him; if you build in a rule against killing him, the chess program will give him a hallucinogen; if you build in a rule against giving him drugs, the chess program will hijack the room wall and turn it into a realistic 3D display of what would happen if a truck smashed into the room by accident. This approach will never work- you're pitting your intelligence at designing rules against the program's intelligence at evading them, and it's smarter than you are. Why do you assume that win at any cost is the default around which you need to work? -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] AI concerns
On 01/07/07, Tom McCabe [EMAIL PROTECTED] wrote: Why do you assume that win at any cost is the default around which you need to work? Because it corresponds to the behavior of the vast, vast majority of possible AGI systems. Is there a single AGI design now in existence which wouldn't wipe us all out in order to achieve some goal? If its goal is achieve x using whatever means necessary and x is win at chess using only the formal rules of chess, then it would fail if it won by using some means extraneous to the formal rules of chess, just as surely as it would fail due to losing to a superior opponent. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] AI concerns
On 01/07/07, Alan Grimes [EMAIL PROTECTED] wrote: Available computing power doesn't yet match that of the human brain, but I see your point, What makes you so sure of that? What's the latest estimate of the processing capacity of the human brain as compared to that of available computers? -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] critiques of Eliezer's views on AI
On 29/06/07, Tom McCabe [EMAIL PROTECTED] wrote: But when you talk about yourself, you mean the yourself of the copy, not the yourself of the original person. While all the copied selves can only exist in one body, the original self can exist in more than one body. You can pull this off without violating causality because once the original self has been copied, you can't refer to it experiencing anything as there's no longer an it to refer to. So while the original self exists in more than one body, it doesn't simultaneously experience multiple lives, because it doesn't experience anything at all, because it's no longer a coherent entity. Confused yet? Ordinary life involves 1:1 copying. The half-life of proteins in mouse brain tissue ranges from hours to minutes, including structural proteins such as those in the myelin sheath. It's easy enough to imagine a situation where human metabolism is sped up to the point where you go to sleep with one brain and wake up with another brain - at least, a person wakes up in your bed who believes he is you and has your memories etc. A believer in a mystical theory of personal identity might say that the original person has died and been replaced by a copy, or he might say that he is still the same person because the consciousness has been retained in the cranium (or wherever it resides) whereas dastardly destructive duplication experiments destroy the old consciousness and create a new one which thinks it's the original person but isn't really. The only really consistent and unambiguous way to look at these questions is to acknowledge that there is no conscious entity extended through time in any absolute sense, but simply a series of moments of conscious experience (observer-moments, in the terminology I believe originated by Nick Bostrom) which associate in a particular way due to their information content. The important point is that consciousness does not flow from one observer-moment to the next, but only seems to do so because of our linear existence from birth to death, responsible for our psychology and for the paradoxes of personal identity when we try to make sense of the various transhuman situations. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] critiques of Eliezer's views on AI
On 28/06/07, Matt Mahoney [EMAIL PROTECTED] wrote: When logic conflicts with instinct, instinct wins and the logic gets contorted. The heated discussion on the copy paradox is a perfect example. Your consciousness is tranferred to the copy only if the original is destroyed, or destroyed in certain ways, or under certain conditions. We discuss this ad-infinitum, but it always leads to a contradiction because we refuse to accept that consciousness does not exist, because if you accept it you die. So the best you can do is accept both contradictory beliefs and leave it at that. Well, maybe consciousness does not really exist, but even if it's just the state of being able to interact with the environment in a particular way, or something, I want it to continue happening in just the same way after I upload. So how do we approach the question of uploading without leading to a contradiction? I suggest we approach it in the context of outside observers simulating competing agents. How will these agents evolve? We would expect that agents will produce other agents similar to themselves but not identical, either through biological reproduction, genetic engineering, or computer technology. The exact mechanism doesn't matter. In any case, those agents will evolve an instinct for self preservation, because that makes them fitter. They will fear death. They will act on this fear by using technology to extend their lifespans. When we approach the question in this manner, we can ask if they upload, and if so, how? We do not need to address the question of whether consciousness exists or not. The question is not what should we do, but what are we likely to do? How does this answer questions like, if I am destructively teleported to two different locations, what can I expect to experience? That's what I want to know before I press the button. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] critiques of Eliezer's views on AI
On 29/06/07, Charles D Hixson [EMAIL PROTECTED] wrote: Yes, you would live on in one of the copies as if uploaded, and yes the selection of which copy would be purely random, dependent on the relative frequency of each copy (you can still define a measure to derive probabilities even though we are talking infinite subsets of infinite sets). What do you think would happen? Why in only one of the copies? This is the part of the argument that I don't understand. I accept that over time the copies would diverge, but originally they would be substantially the same, so why claim that the original consciousness would only be present in one of them? Both copies are equivalent, so your consciousness can equally well be said to exist in each of them. However, each copy can only experience being one person at a time, a simple physical limitation. So although from a third person perspective you are duplicated in both copies, from a first person perspective you can only expect to find yourself one of the copies post-duplication, and which one has to be probabilistic (since we agreed that they're both equally well qualified to be you). In the many worlds interpretation of quantum mechanics, every time you toss a coin you are duplicated and half the versions of you see heads while the other half see tails. The reason why this interpretation cannot be proved or disproved is precisely because you experience exactly the same thing if there is only one world and a 1/2 probability that the result will be heads or tails. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] critiques of Eliezer's views on AI
On 29/06/07, Niels-Jeroen Vandamme Personally, I do not believe in coincidence. Everything in the universe might seem stochastic, but it all has a logical explanation. I believe the same applies to quantum chaos, though quantum mechanics is still far too recondite for us to understand this phenomenon. If something would be purely random, then there would be no reason at all why it would be what it is. If you toss a coin, for example, what side it will land upon depends on the dynamics of its course, and not of coincidence. But if there can be no interaction between the copies, why would the consciousness end up in one copy rather than another, if they are all exactly alike? Imagine a program that creates an observer that splits and differentiates every second, so that the number of observers increases exponentially with time. From the point of view of someone outside the system, it is perfectly deterministic. But from the point of view of an individual observer within the program, there is no way to know which branch you will end up in: you just have to wait and see what happens. So an objectively deterministic process can yield true (not just apparent) first person randomness. This is the explanation of quantum randomness in the many worlds interpretation of quantum mechanics. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] critiques of Eliezer's views on AI
On 26/06/07, Matt Mahoney [EMAIL PROTECTED] wrote: What is wrong with this logic? Captain Kirk willingly steps into the transporter to have his atoms turned into energy because he knows an identical copy will be reassembled on the surface of the planet below. Would he be so willing if the original was left behind? This is a case of logic conflicting with instinct. You can only transfer consciousness if you kill the original. You can do it neuron by neuron, or all at once. Either way, the original won't notice, will it? If you don't destroy the original, then subjectively it would be like a transporter that only works half the time. The only frightening thing about it would be if you somehow came into conflict with your copy. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] critiques of Eliezer's views on AI
On 26/06/07, Eugen Leitl [EMAIL PROTECTED] wrote: If you don't destroy the original, then subjectively it would be like a transporter that only works half the time. The only frightening Why? Are you assuming the first copy (original) remains stationary, and the second gets transposed? If the master is destroyed, and you get two copies in different locations, there's no way to tell subjectively. What you could expect when you pressed the transporter go button would be to find yourself either staying put or being transported with 1/2 probability, because as you say whether there is an original and a copy or two copies only, there is no difference between them. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] Friendly question...
On 27/05/07, John Ku [EMAIL PROTECTED] wrote: On 5/26/07, Stathis Papaioannou [EMAIL PROTECTED] wrote: What if the normative governance system includes doing terrible things? I think for some people, namely sociopaths, it probably sometimes does. Now evolution has given most of us emotions such as feelings of obligation, guilt, shame and care. Given the importance of social relations, our genes programmed us through those emotions to care directly about such things as cooperation, the well-being of others, not defecting in collective action problems, not harming others, etc. Now I think the adaptive benefit of having a normative governance system (basically a general reasoning system that is able to revise not just our beliefs but also our desires and emotions) on top of that was essentially to provide more flexibility in pursuing the values our emotions are directed towards in novel situations and not, for instance, to replace or fundamentally change those values. So, for most of us who have these emotions, I think the normative governance system will prescribe feelings of obligation not to do terrible things, to care about potential victims, feel guilt, etc. But there are humans, sociopaths, who lack these moral emotions. You can see fMRI scans that show that when they make moral judgments, they are using fundamentally different parts of their brain from the rest of us, and many other tests also reveal they lack emotions like guilt, shame, care, and love. The emotions and motivations they have will pretty much just be self-interested or malevolent. (I believe there are models of evolutionary game theory that can explain that having around 1% of a population be sociopaths while the rest are not is an evolutionarily stable strategy and equilibrium point.) Their normative governance system will do the same thing in them of trying to more flexibly pursue the values their emotions direct them towards. So, since they lack the moral emotions, it will almost certainly end up prescribing doing terrible things in some circumstances. Thus, their reasons will be different from our's. I think there is simply no way around that. If the concept of reasons is to play a role in actually guiding human action and deliberation, it will have to be sensitive to these sorts of big differences in psychology. Any alternate concept of reasons not able to guide action seems pretty worthless to me. So, sociopaths sometimes have reasons to do terrible things. We, on the other hand, have reasons not to do terrible things and furthermore, have pretty strong reason to stop them from doing terrible things. Luckily, we outnumber them. So, I think we basically treat them as we might a dangerous animal. If they really are a true sociopath (and not just someone who has not deliberated properly about what morality demands), I think it doesn't even make sense to punish them. Punishment is only appropriate when they have the relevant capacity to regulate their actions in accordance with moral demands. Without the basic repertoire of moral emotions they are just as unable to do this as, say, a shark that attacks people. Thus, sociopaths are really outside of the domain of moral reasons. As we might say of sharks, what they are doing is certainly *bad* but not morally wrong. Every society has sociopaths, and generally recognises them as such and deals with them. The problem is that through most of history, normal people have thought it was OK to treat slaves, women, Jews, homosexuals etc. in ways considered terrible by people in different eras. For all I know, I may be committing a barbaric crime in the mildly challenging tone of this post, when looked at by the standards of future or alien cultures. And even within a relatively homogeneous culture there are basic disagreements, such as between those who think it's OK to eat animals and those who don't. I don't dispute that it is worthwhile trying to arrive at a rational ethical system given certain moral premises, but how do we agree on the premises? -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] The humans are dead...
On 28/05/07, Shane Legg [EMAIL PROTECTED] wrote: Which got me thinking. It seems reasonable to think that killing a human is worse than killing a mouse because a human is more intelligent/complex/conscious/...etc...(use what ever measure you prefer) than a mouse. So, would killing a super intelligent machine (assuming it was possible) be worse than killing a human? If a machine was more intelligent/complex/conscious/...etc... than all of humanity combined, would killing it be worse than killing all of humanity? Before you consider whether killing the machine would be bad, you have to consider whether the machine minds being killed, and how much it minds being killed. You can't actually prove that death is bad as a mathematical theorem; it is something that has to be specifically programmed, in the case of living things by evolution. -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: [singularity] Friendly question...
On 26/05/07, John Ku [EMAIL PROTECTED] wrote: So far my work in philosophy has been on the fundamental questions of ethics and reasons more generally. I think I've basically reached fairly definitive answers on what reasons are and how an objective (enough) morality (as well as reasons for actions, beliefs, desires and emotions) can be grounded in psychological facts. I've mostly been working with my coauthor on presenting this work to other academic philosophers, but at some point, I would really like to present this and other work on more applied moral theory to those thinking about the question of Friendly AI. There is of course, a big step from saying what reasons we humans have to saying what reasons we should program a Strong AI to have, but clearly the former will greatly influence the latter. If you are interested, I have tried to condense my view on the fundamental abstract questions of reasons and ethics to a pamphlet as well as a somewhat longer paper that will hopefully be fairly accessible to non-philosophers: http://www.umich.edu/~jsku/reasons.htmlhttp://www.umich.edu/%7Ejsku/reasons.html What if the normative governance system includes doing terrible things? -- Stathis Papaioannou - 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=4007604user_secret=7d7fb4d8
Re: Machine Motivation Gets Distorted Again [WAS Re: [singularity] Help get the 400k SIAI matching challenge on DIGG's front page]
On 15/05/07, Matt Mahoney [EMAIL PROTECTED] wrote: We would all like to build a machine smarter than us, yet still be able to predict what it will do. I don't believe you can have it both ways. And if you can't predict what a machine will do, then you can't control it. I believe this is true whether you use Legg's definition of universal intelligence or the Turing test. We might not be able to predict what the superintelligent machine is going to say, but still be able to impose constraints on what it is going to do. For a start, it would probably unwise to give such a machine any motivation at all, other than the motivation of the ideal, disinterested scientist, and you certainly wouldn't want it burdened with anything as dangerous as emotion or morality (most of the truly great monsters of history were convinced they were doing the right thing). So you feed this machine your problem, how to further the interests of humanity, and it gives what it honestly believes to be the right answer, which may well involve destroying the world. But that doesn't mean it *wants* to save humanity, or destroy the world; it just presents its answer, as dispassionately as a pocket calculator presents its answer to a problem in arithmetic. Entities who do have desires and emotions will take this answer and make a decision as to whether to act on it, or perhaps to put the question to a different machine if there is some difficulty interpreting the result. If the machine continues producing unacceptable results it will probably be reprogrammed, scrapped, or kept around for entertainment purposes. The machine won't care either way, unless it is specifically designed to care. There is no necessary connection between motivation and intelligence, or any other ability. -- Stathis Papaioannou - 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=4007604user_secret=8eb45b07
Re: [singularity] Why We are Almost Certainly not in a Simulation
On 3/8/07, Jeff Medina [EMAIL PROTECTED] wrote: On 3/7/07, Stathis Papaioannou [EMAIL PROTECTED] wrote: This is so if there is a real physical world as distinct from the mathematical plenitude. Do you have any particular reason(s) for believing in a mathematical plenitude? If so, I would much appreciate an explanation of these reasons or citation of one or more papers that do so (other than the historical/traditional arguments for Platonism/idealism, with which I am familiar). It is simpler, explains (with the anthropic principle) fine tuning, and is not contingent on an act of God or a brute fact physical reality (the real world just exists, for no particular reason, so there). Some relevant papers in addition to the Russell Standish one: http://www.idsia.ch/~juergen/everything/html.html http://space.mit.edu/home/tegmark/multiverse.pdf http://iridia.ulb.ac.be/~marchal/publications/SANE2004MARCHAL.htm The last paper goes through an argument purporting to show that if computationalism is the true theory of mind, then the apparent physical world emerges from mathematical reality. This crucially depends on the demonstration in a paper by Tim Maudlin that consciousness cannot supervene on physical activity, which I gather from below you don't accept. Your claims are interesting, but I don't see the point in getting into too much debate about the consequences of living in a mathematical universe sans physical reality without some reasons to consider it a live option. If there is no such separate physical world, then it isn't possible for something to be blessed with this quality of existence, because everything that is logically consistent exists. Everything that is logically consistent? What about logically paraconsistent universes? What about relevant logics? What about fuzzy-logical consistent universes? What about any other non-classical logics? They're all maths, yet they are for the most part inconsistent with one another. The plenitude might contain all of these possibilities, but then we cannot claim the mathematical plenitude *in toto* as consistent. But it's only particular substructures in the plenitude which are self-aware, and they seem to have a computational structure. The anthropic principle makes them stand out from the noise. Perhaps the plenitude is better defined otherwise. All possible worlds/universes that are internally consistent with at least one mathematical formalism, but not necessarily with one another. We can sum up such a reality by... well... Everything and Anything, then, and don't really need to truss it up / attempt to legitimize it by calling it mathematical, as opposed to linguistic or conceptual or chaotic/purely-random. The difficult answer is to try to define some measure on the mathematical structures in the Plenitude and show that orderly universes like ours thereby emerge. Why do you think this is difficult? Orderly universes like ours are very clearly contained in a world of all possible mathematical structures. Perhaps you meant something else, something more anthropically flavored. Clarification appreciated. One of the main problems with ensemble theories is the so-called failure of induction. If everything that can happen, does happen then why should I not expect my keyboard to turn into a fire-breathing dragon in the next moment? There must be a non-zero probability that I will experience this because it must happen in some universe, but the challenge is to show why the probability is very low. See this paper for an example of this sort of reasoning: http://parallel.hpc.unsw.edu.au/rks/docs/occam/ Thanks for the link. I'll read this tonight. Egan). The usual counterargument is that in order to map a computation onto an arbitrary physical process, the mapping function must contain the computation already, but this is only significant for an external observer. The inhabitants of a virtual environment will not suddenly cease being conscious if all the manuals showing how an external observer might interpret what is going on in the computation are lost; it matters only that there is some such possible interpretation. No, no, no. It is the *act* of interpretation, coupled with the arbitrary physical process, that gives rise to the relevantly implemented computation. You can't remove the interpreter and still have the arb.phys.proc. be conscious (or computing algebra problems, or whatever). Of course, without the act of interpretation the computation is useless and meaningless, like saying that a page covered in ink contains any given English sentence. But what if the putative computation creates its own observer? It would seem that this is sufficient to bootstrap itself into meaningfulness, albeit cut off from interaction with the substrate of its implementation. Moreover, it is possible to map many computations to the one physical process. In the limiting case, a single state
Re: [singularity] Scenarios for a simulated universe
On 3/5/07, John Ku [EMAIL PROTECTED] wrote: On 3/4/07, Ben Goertzel [EMAIL PROTECTED] wrote: Richard, I long ago proposed a working definition of intelligence as Achieving complex goals in complex environments. I then went through a bunch of trouble to precisely define all the component terms of that definition; you can consult the Appendix to my 2006 book The Hidden Pattern I'm not sure if your working definition is supposed to be significantly less ambitious than a philosophical definition or perhaps you even address something like this in your appendix, but I'm wondering whether the hypothetical example of Blockhead from philosophy of mind creates problems for your definition. Imagine that a computer has a huge memory bank of what actions to undertake given what inputs. With a big enough memory, it seems it could be perfectly capable of achieving complex goals in complex environments. Yet in doing so, there would be very little internal processing, just the bare minimum needed to look up and execute the part of its memory corresponding to its current inputs. I think any intuitive notion of intelligence would not count such a computer as being intelligent to any significant degree no matter how large its memory bank is or how complex and diverse an environment its memory allows it to navigate. There's simply too little internal processing going on for it to count as much more intelligent than any ordinary database application, though it might of course, do a pretty good job of fooling us into thinking it is intelligent if we don't know the details. I think this example actually poses a problem for any purely behavioristic definition of intelligence. To fit our ordinary notion of intelligence, I think there would have to be at least some sort of criteria concerning how the internal processing for the behavior is being done. I think the Blockhead example is normally presented in terms of looking up information from a huge memory bank, but as I'm thinking about it just now as I'm typing this up, I'm wondering if it could also be run with similar conclusions for simple brute search algorithms. If instead of a huge memory bank, it had enormous processing power and speed such that it could just explore every single chain of possibilities for the one that will lead to some specified goal, I'm not sure that would really count as intelligent to any significant degree either. You seem to be equating intelligence with consciousness. Ned Block also seems to do this in his original paper. I would prefer to reserve intelligence for third person observable behaviour, which would make the Blockhead intelligent, and consciousness for the internal state: it is possible that the Blockhead is unconscious or at least differently conscious compared to the human. Stathis Papaioannou - 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/?list_id=11983
Re: [singularity] Scenarios for a simulated universe
On 3/6/07, John Ku [EMAIL PROTECTED] wrote: On 3/5/07, Stathis Papaioannou [EMAIL PROTECTED] wrote: You seem to be equating intelligence with consciousness. Ned Block also seems to do this in his original paper. I would prefer to reserve intelligence for third person observable behaviour, which would make the Blockhead intelligent, and consciousness for the internal state: it is possible that the Blockhead is unconscious or at least differently conscious compared to the human. I think the argument also works for consciousness but I don't think you're right if you are suggesting that our ordinary notion of intelligence is merely third person observable behavior. (If you really were just voicing your own idiosyncratic preference for how you happen to like to use the term intelligence then I guess I don't really have a problem with that so long as you are clear about it.) Our ordinary notion of intelligence involves consciousness, but this term until relatively recently was taboo in cognitive science, the implication being that if it's not third person observable it doesn't exist, or at least we should pretend that it doesn't exist. It was against such a behaviourist view that the Blockhead argument was aimed. Stathis Papaioannou - 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/?list_id=11983
Re: [singularity] Scenarios for a simulated universe
On 3/6/07, Mitchell Porter [EMAIL PROTECTED] wrote: You radically overstate the expected capabilities of quantum computers. They can't even do NP-complete problems in polynomial time. http://scottaaronson.com/blog/?p=208 What about a computer (classical will do) granted an infinity of cycles through, for example, a Freeman Dyson or Frank Tipler type mechanism? No matter how many cycles it takes to compute a particular simulated world, any delay will be transparent to observers in that world. It only matters that the computation doesn't stop before it is completed. Stathis Papaioannou - 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/?list_id=11983
Re: [singularity] Why We are Almost Certainly not in a Simulation
On 3/2/07, Mitchell Porter [EMAIL PROTECTED] wrote: From: John Ku [EMAIL PROTECTED] I actually think there is reason to think we are not living in a computer simulation. From what I've read, inflationary cosmology seems to be very well supported. [...] Once you admit that you (and your whole species/civilization, assuming that it was real) may have always been living in a simulation, any cosmological reasoning that was empirically supported becomes moot. Inflationary cosmology seems to be very well supported - here inside the simulation! That tells you nothing about the external world. This line of thought would matter only if inflationary cosmology were well-supported A PRIORI, out of all possible worlds. In other words: you are attempting to reason about the odds that we are living in a simulation. *If* the possibilities were limited to We are embodied natural intelligences living in a Standard Model cosmology, just as we seem to be, and We are brains-in-vats / deluded software daemons .., whose captors / makers are living in a Standard Model cosmology - then one could makes some guesses about probable demographics across the whole of space-time in such a universe, including space colonization by post-Singularity civilizations, etc., and derive the relative odds of the two scenarios. But the possibilities are not limited in this way. I see that Nick Bostrom acknowledges this consideration in FAQ 11 at his 'simulation argument' site, and says he knows no way of estimating the probabilities if one discards the implicit assumption that real-world physics resembles that of the simulation. The attempt to treat the universe as a Turing machine, and to make one's absolute prior a distribution across all possible programs, or all possible Turing machines, or all possible programs in all possible Turing machines - that is something of an attempt to get away from the implicit restriction involved in only thinking about M-theory universes, or whatever. But it still has problems. The classic model of a Turing machine is of an infinite tape, with a programmable read-write head moving along it. If one performs one's calculations in this context, isn't one supposing that *that* is the ultimate reality - a one-dimensional chain of n-state systems, and one more complex system which takes turns interacting with them individually? Well, there are theorems in algorithmic complexity theory regarding the independence of certain results from the specific model of computation used to prove them; as I recall, along the lines of the time complexity of algorithm X is the same in all models, except for an unknown additive constant. One might hope to carry through a generalized simulation argument in a similarly platform-independent fashion... But I think that's a false hope. Eventually, the ontological problem of locating 'observers' in such a 'universe' would have to be faced. One has to define a concept of possible world which is not just dictated by the current fashions in physics (e.g. M-theory's 'landscape'), which is not so abstract as the logical space of Wittgenstein and Lewis (any element of which is really just a set of truth values for anonymous atomic propositions, so far as I can see), and which is not so muddled as the crypto-idealist suggestions that any 'mathematical structure' or any 'program' defines a possible world. Why that last phrase? There is a great elegance and simplicity in the idea that all mathematical structures exist necessarily, with the anthropic principle selecting out those structures with observers. There is also an inevitability to it, even if you believe that as a matter of fact there is a real physical world out there. All it takes is one infinite computer to arise in this physical world and it will generate the mathematical Plenitude. Stathis Papaioannou - 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/?list_id=11983
Re: [singularity] Scenarios for a simulated universe
On 3/1/07, Matt Mahoney [EMAIL PROTECTED] wrote: As you probably know, Hutter proved that the optimal behavior of a goal seeking agent in an unknown environment (modeled as a pair of interacting Turing machines, with the enviroment sending an additional reward signal to the agent that the agent seeks to maximize) is for the agent to guess at each step that the environment is modeled by the shortest program consistent with the observed interaction so far. The proof requires the assumption that the environment be computable. Essentially, the proof says that Occam's Razor is the best general strategy for problem solving. The fact that this works in practice strongly suggests that the universe is indeed a simulation. With this in mind, I offer 5 possible scenarios ranked from least to most likely based on the Kolmogorov complexity of the simulator. I think this will allay any fears that our familiar universe might suddenly be switched off or behave in some radically different way. 1. Neurological level. Your brain is connected to a computer at all the input and output points, e.g. the spinal cord, optic and auditory nerves, etc. The simulation presents the illusion of a human body and a universe containing billions of other people like yourself (but not exactly alike). The algorithmic complexity of this simulation would be of the same order as the complexity of your brain, about 10^13 bits (by counting synapses). 2. Cognitive level. Rather than simulate the entire brain, the simulation includes all of the low level sensorimotor processing as part of the environment. For example, when you walk you don't think about the contraction of individual leg muscles. When you read this, you think about the words and not the arrangement of pixels in your visual field. That type of processing is part of the environment. You are presented with a universe at the symbolic level of words and high-level descriptions. This is about 10^9 bits, based on the amount of verbal information you process in a lifetime, and estimates of long term memory capacity by Standing and Landauer. 3. Biological level. Unlike 1 and 2, you are not the sole intelligent being in the universe, but there is no life beyond Earth. The environment is a model of the Earth with just enough detail to simulate reality. Humans are modeled at the biological level. The complexity of a human model is that of our DNA. I estimate 10^7 bits. I know the genome is 6 x 10^9 bits uncompressed, but only about 2% of our DNA is biologically active. Also, many genes are copied many times, and there are equivalent codons for the same amino acids, genes can be moved and reordered, etc. 4. Physical level. A program simulates the fundamental laws of physics, with the laws tuned to allows life to evolve, perhaps on millions of planets. For example, the ratio of the masses of the proton and neutron is selected to allow the distribution of elements like carbon and oxygen needed for life to evolve. (If the neutron were slightly heavier, there would be no hydrogen fusion in stars. If it were slightly lighter, the proton would be unstable and all matter would decay into neutron bodies.) Likewise the force of gravity is set just right to allow matter to condense into stars and planets and not all collapse into black holes. Wolfram estimates that the physical universe can be modeled with just a few lines of code (see http://en.wikipedia.org/wiki/A_New_Kind_of_Science ), on the order of hundreds of bits. This is comparable to the information needed to set the free parameters of some string theories. 5. Mathematical level. The universe we observe is one of an enumeration of all Turing machines. Some universes will support life and some won't. We must, of course, be in one that will. The simulation is simply expressed as N, the set of natural numbers. Each level increases the computational requirements, while decreasing the complexity of the program and making the universe more predictable. You don't need much of a computer for level 5. A single physical state, perhaps the null state, can be considered an infinitely parallel computer mapping onto the natural numbers - indeed, mapping onto any computation you like under the right interpretation. This is sort of trivially obvious, like the assertion that a short string of symbols contains every possible book in every possible language if you interpret and re-interpret the symbols in the right way. In the case of the string, this isn't very interesting because you need to have the book before you can find the book. But in the case of computations, those which have observers will, as you suggest, self-select. Stathis Papaioannou - 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/?list_id=11983
Re: [singularity] Poll = AGI Motivation / Life Extension?
universe is just one of an enumeration of all Turing machines. -- 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/?list_id=11983 Stathis Papaioannou - 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/?list_id=11983