On Sat, Jul 9, 2011 at 12:07 AM, Colin Geoffrey Hales < cgha...@unimelb.edu.au> wrote:
> ** > > Down the bottom.... if you dare.... there be dragons... :-) > > > -----Original Message----- > From: email@example.com on behalf of Jason Resch > Sent: Sat 7/9/2011 1:23 AM > To: firstname.lastname@example.org > Subject: Re: COMP refutation paper - finally out > > On Fri, Jul 8, 2011 at 1:56 AM, Colin Geoffrey Hales < > cgha...@unimelb.edu.au > > wrote: > > > Hi,**** > > > > **** > > > > You have missed the point. When you feel pain in your hand your are > feeling > > it because the physics of specific specialized small regions of the > cranial > > central nervous system are doing things. This includes (1) action > potentials > > mutually resonating with (2) a gigantic EM field system in extremely > complex > > ways. *Exactly how and why this specific arrangement of atoms and > > behaviour delivers it is irrelevant. It is enough to know that it does*. > > More than that it is the ONLY example of natural cognition we have.**** > > > > **** > > > > The whole point of this argument is that unlike any other time in the > > history of science, we are expecting the particular physics (that we know > > delivers cognition) can be totally replaced (by the physics of a > computer > > or even worse, a non-existent Turing machine) , yet still result in > > cognition. **** > > > > > > It's not the "totally" that is the problem. Bruno asks if you can > replace > > a part of a brain with something that does the same computation (at some > > level) and have no effect on the conscious (or unconscious) life of that > > person. This certainly seems plausible. But it relies on the remaining > > world to continue interacting with that person. So in his idea of > replacing > > physics with computation he has to suppose replacing all of the brain > plus > > everything that interacts with the brain. In other words a simulation of > > the person(s) and the universe. Then within the simulation EM fields are > > computed and supply computed illumination to computed eyes and brains. > He > > invites us to consider all this computation done by a universal > dovetailer, > > a computer which also computes all possible computable universes as it > > goes. But to me it seems a great leap from computing what a piece (or > even > > all) of a brain does to computing a whole (quantum) universe. I'm not at > > all sure that the universe is computable; and it's certainly a different > > question than whether I would say yes to the doctor. > > > > *This entire scenario has nothing to do with what I am talking about. > > Bruno is talking about the universe AS abstract computation. Ontology. I > am > > talking about a completely different area: the computation of > descriptions > > of a universe; descriptions compiled by observers within it called > 'laws > > of nature'. *** > > > > ** ** > > > > *This is the main problem. We are speaking at cross purposes. Computation > > by computers made of bits of our universe is not the same is describing > of a > > universe of ontological primitives interacting. I find the latter really > > interesting, but completely irrelevant to the task at hand, which is to > > create artificial cognition using the real world of humans and the stuff > > they are made of. * > > > > **** > > > > If you believe that computed physics equations is indistinguishable from > > physics, to the point that a computed model of the physics of cognition > is > > cognition, then why don't you expect a computed model of combustion > physics > > to burst into flames and replace your cooker? Why can't you go to work in > a > > computed model of a car that spontaneously springs into your life? Why > don't > > you expect to be able to light your room with a computed model of the > > physics of a lightbulb? Why can't you compute Maxwell's equations and > create > > a power station?**** > > > > > > You can within a simulation.**** > > > > ** ** > > > > *At last, someone takes the magical step. This is the problem writ-large. > > What you are saying, in effect, is that computation about X is only some > > kind of simulation of X. My whole point is that I do not want a > simulation > > of X. I want an X. Like artificial fire is still fire. Like artificial > light > > is light. Like artificial lightning is lightning. Like artificial > cognition > > is cognition. Like an artificial round rollything (wheel) is a wheel. > .... > > like a million other artificial versions of a natural phenomenon created > by > > humans for millennia.* > > > > * * > > > > *In using a computer, all the original physics is gone. Yet the 100% > > expectation is (apart from yourself, apparently... or.not... we have > found > > the inconsistency at last) that computers will lead to AGI is the state > of > > the game. Yet it involves entirely disposing of the natural phenomenon > that > > we know originates it. It replaces the entire physics with the physics of > a > > computer ... and then expects to get the natural phenomenon out of it!* > > > > Ultimately physics is just set of well defined rules (algorithms) and > matter and energy is just information. > > It is not unreasonable to imagine that information sits at the core of > > physics, just as it sits at the core of a computer. > > > > It from bit. Otherwise put, every 'it'-every particle, every field of > > force, even the space-time continuum itself-derives its function, its > > meaning, its very existence entirely-even if in some contexts > > indirectly-from the apparatus-elicited answers to yes-or-no questions, > > binary choices, bits. 'It from bit' symbolizes the idea that every item > of > > the physical world has at bottom-a very deep bottom, in most instances-an > > immaterial source and explanation; that which we call reality arises in > the > > last analysis from the posing of yes-no questions and the registering of > > equipment-evoked responses; in short, that all things physical are > > information-theoretic in origin and that this is a participatory > universe. > > (John Archibald Wheeler 1990: 5) > > > > More and more information is being shown to be a critical property of the > universe, in entropy, the holographic principle, quantum mechanics, > Shanon's > laws of information carrying capacity, extreme physical information etc. > It > is not unreasonable, therefore, to consider the universe has some > informational basis. > > > The ultimate description of anything is information. Information cannot be > described in terms of anything else. You might say the map is not the > territory,. but I would argue that a map that is accurate in every possible > detail is the territory. There would be no way for any observer to > distinguish between the two. You might consider Ockham in the following > thought experiment: Imagine there are two universes, one is the physical > universe in which we live, another is informational representation existing > in math. If observers in the informational version are as conscious as you > and I (I know you would say they are not but stay with me) then they would > also consider their universe to exist physically, there is no way for them > to tell one way or the other. > > Now if an informational universe is shown to exist in math, what is the > reason for postulating our universe is somehow different? In what manner > is > "physical" different from an "informational" one, what does being > "physical" > mean and what does it add which is not already there in the identical > informational one? Everything that happens in the physical universe also > happens in the informational one, they have all the same laws, and the same > evolution through time. What distinguishes the "informational hydrogen > atom" in the informational universe from the "physical hydrogen atom" in > the > physical universe? I imagine you will say something like "It has to be > physical to be real", but I have never seen a convincing argument for this; > perhaps you could provide one? > > Jason > > ============================================================= > Hi again on this fine Saturday when I should be gardening. > > You can 'imagine' whatever you want about the word 'information'. It's > missing the point. > > Here in the _real_ world, where 100% of AGI research is using real > computers (or hardware computation of a model eg FPGA) to implement an AGI. > That is, to implement an actual instance of cognition. The goal is the > equivalent of human cognition. I'll spell it out properly because it seems > to me, after 10 years of intense effort, that I've finally found a simple > achilles heel, where the belief system is shown to be inconsistent. That is, > there is persistence with a belief in X in spite of empirical evidence that > X is false. Like pulling Pinocchio the puppet's strings and then really > believing it creates a real boy. > > ==================== EL PROOF > ===================================================== > 1) We have only 1 benchmark natural general intelligence (NGI): the human. > Dolphins appear to have general intelligence: http://www.guardian.co.uk/science/2003/jul/03/research.science > > 2) Specific physics exists in the brain. The physics of cognition. This is > the physics of action potential propagation (AP) resonating with a large, > unified electromagnetic (EM) field system. The relationship between AP > production to EM field production is _not_ 1:1. More than one EM field is > consistent with one set of AP transmissions. Call it AP/EM physics of > cognition. It's what we have. > Trace the causes from the time someone is shown a picture of a tomato and asked what color it is, until the time the person's lips move to provide an answer. The lip movement is caused by nerve signals sent from the brain, which in turn is sent by other nerve signals from processing within the brain. These signals had their genesis with the sensory input generated by the presentation of the tomato and the question received through the auditory nerves. If one can manufacture an artificial neuron which will have the same inputs and outputs as any other neuron then we can produce one neuron, a thousand, or a billion, and the whole brain will work the same. In previous posts you have indicated that you do not believe in philosophical zombies, so you ought to agree that if it operated the same as the biological brain, it would still be conscious. Now consider that these artificial neurons are not making the decision of when to fire, instead they have wireless transmitters which send the inputs to a nearby computer and the computer replies when the neuron should fire. Again, we have identically behaving individual neurons and therefore an identically behaving brain. The final step is to include a map of all the interconnections of the neurons to the computer, and to then the computer can reproduce the entire system, without any artificial neurons at all. (You could leave some in attached to the sensory inputs and spinal cord outputs, and then it would be able to control a human body in the same way any brain does). What is the flaw in this? Where do the unmodelable causes and effects come into play? Are you still of the opinion that zombies are impossible? > > 3) In the artificial instantiation of fire by humans, the physics of > combustion is retained. In the artificial instantiation of flight by humans, > the physics of flight is retained. And so forth. Without exception. (except > in AGI!). By extension, the way to artificial cognition is to retain the > physics of cognition. AP/EM physics. > Have you ever played around with virtual machines? They allow you to run an instance of a computer as software entirely within a different operating system and hardware. The virtual machine is indistinguishable from a physical machine. At work I have a Windows computer running inside a Linux computer, and at home I have a virtual Linux computer running within a Windows computer. This is an example where a perfect recreation of the laws create an indistinguishable environment for the virtualized system. The windows computer running as a VM has no idea it is running as a software system rather than physical hardware, and in principal it is actually impossible for it to determine on what hardware it is ultimately running. Now if I created a virtual physical system, what clue would that have or could they have that they are running as software rather than a physical platform? They would behave identically, just as my virtual windows machine behaves identically to a physical windows system. Given your disbelieve in zombies, you must accept it would be a mind in all the same respects as a physical incarnation of a mind. > > 4) If computing can literally duplicate human cognition, then by extension, > computing the laws of combustion should create flames. REAL FLAMES. That is > the logical result of a belief that computing the physics (or worse, an > abstraction of it) of cognition leads to NGI cognition. > If you dreampt of fire would you expect your head to burst into flames? > > 5) The computation of the physics of flames has been computed _to death_ > for decades. NO FLAMES resulted. In general, no instance of computation of a > physics model of X ever resulted in X. In a very Popperian way, we have > experimentally already falsified, decades ago, a hypothesis that that > computation can produce human cognition. > I could counter that the simulation of the hardware running a windows operating system results in a windows operating system. Thus we have experimentally confirmed the hypothesis that computation can produce human cognition. > > Therefore, > > 6) A claim that a computer can deliver human cognition is false. QED. > > > To continue to believe that a Turing-based computation of a model of > cognition is cognition is thus empirically falsified and has been for a long > time. This is GAME OVER for purely computational AGI. > I think over the next few decades your position will be shown as incorrect when we construct models of human brains that behave and respond in the same way we expect their biological counterparts to. This has already been done for smaller sections of the brain, such as the neocortical column, which is a building block of the mamallian brain. See: https://secure.wikimedia.org/wikipedia/en/wiki/Blue_Brain_Project and http://news.bbc.co.uk/2/hi/8164060.stm > Not in a formal sense, but in a technological investment sense consistent > with every other instance of technological instantiation of a natural > phenomenon. What sort of idiot would you have to be to invest in a new kind > of match, where you try to light your cigarette with a computation of the > physics of combustion ... because that is exactly what investment in > computational AGI is doing. > When I play Grand Theft Auto, I can set things on fire within the game using the in-game flame thrower. This causes the in-game people to react with fear and run away. Already games include AI models with some level of intelligence (fire is dangerous, avoid it). What is preventing game designers from making smarter and smarter in-game entities to interact with? You agree that there are computer programs that can play chess better than most humans, and driving software that can drive better than most humans (Google cars logged over 140,000 miles with no accidents). If humans implemented a billion such programs involving different skills and functions, and put them all into the same computer, enabling it to competently do whatever any person can do, would you call this computer generally intelligent? Why do you think a computer can play chess but not be intelligent "generally"? How do you define "general intelligence"? Jason -- You received this message because you are subscribed to the Google Groups "Everything List" group. To post to this group, send email to email@example.com. To unsubscribe from this group, send email to everything-list+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/everything-list?hl=en.