On Wed, Jul 6, 2011 at 10:22 PM, Colin Geoffrey Hales < cgha...@unimelb.edu.au> wrote:
> Hi Richard et. al., > Wow that thread just keeps on going! > > I am designing chips that do what the brain does. There is ZERO > computing. The use of the chips is, I believe a viable source of > empirical verification of the claims of the kind that have been > discussed in this thread, insofar as any practical outcomes speak to > that end. Rather than talk about it, I'd rather just build them and > start using them. > > I'd like to make an interesting (at least I think, anyway) observation. > > AGI is a case of technology design, unlike any other design in the > history of technology engineering, in the construction of an artificial > version of a natural phenomenon, the original physics of the natural > phenomenon (the activity found in a brain) has been discarded, yet the > natural phenomenon is expected to occur! > The telephone enables conversations to occur over a digital, informational medium, discarding the original physics of the direct vibration from one person's mouth to the other's ear. With the appropriate wiring, sound and vibration could be eliminated altogether, using sensors in the throat muscles to determine speech ( http://www.cerebralpalsylawdoctor.com/technology/sensor-based-system-could-help-sufferers-of-cerebral-palsy.html) and replacing the headphone with wires connected directly to one's cochlea. When it comes to information transmission, storage, processing, there are many physical substitutes and possible instantiations. > > Consider if such an approach had been adopted in the case of the > creation of 'artificial fire'. Gronk the ancient hominid has pushed a > rock into a clearing and carefully painted flames on the side of the > rock. Gronk then erects meat on a spit over the pretend flames and waits > for the feast to be cooked. The physics of fire has been replaced by the > physics of a rock. In the modern approach to AGI, in exactly the same > way, the physics of cognition has been replaced by the physics of a > slightly more complicated, hotter rock: a semiconductor chipset. Instead > of the flames drawn on the rock, we have drawn the program in the > chipset. Just like Gronk we await our meal of AGI. After 60 years, we > are still waiting. What delusion has us believing it ever will? > What Gronk wanted was the experience of a warm meal. This can be achieved without the need for any meal at all, as our dreams show is possible, and in the same way a flight simulator can provide the experience of flight without leaving the ground. What leads you to believe the material in the brain must contain only particals having certain chemical or nuclear bonds? In the end, that is the only difference between the material in the brain and the material in a rock or a computer chip. After all, they are all made of quarks and electrons. Saying thought is only possible with certain biological material is like saying flight is only possible with feathers. With flight, it is the design and organization of the matter which is important to getting off the ground. The same is claimed by computationalism regarding thought. Also, to expect AGI with the machines of today, (never mind those of 60 years ago) is like expecting a pocket calculator to render the movie Avatar. I've estimated that just to store the connections of all the synapses in the human brain would require about 50,000 TB (nevermind the processing that would be required). Our technology is, however, quickly closing in on these levels of storage and processing. With the current rate of growth these levels will be met within a few decades. Any regular user of the Internet knows how difficult it is to read CAPTCHA's, I struggle to correctly enter them all the time, take a look at Google's: https://www.google.com/accounts/DisplayUnlockCaptcha They are so difficult today because AI capable of recognizing text is that advanced. As AI advances in little steps the domain of what we consider requires *true *intelligence is continually shrinking. If you had a time machine to go back to the 1970s and you told them everything AI has accomplished they would think we were most of the way there, with our voice recognition, automatic translation, self driving cars, automated credit card fraud detection and classification of commercial junk e-mail, artificial grandmaster chess players, and so on. These are tremendous accomplishments, and that its taken 60 years to get this far is nothing to scoff at when you consider it took nature billions of years to get to us. General intelligence is just a very big collection of specialized intelligences, it is not a binary thing that you either have or do not (in my opinion). > > You can list example after example of the implementation of artificial > human technology that can be expressed in the same way (eg flight, the > wheel, telephone, transport, power generation, lightning, the lightbulb > ..etc etc). In some technology, computation has been very successfully > included in the causal chain that implements the technology. However, in > terms of the artificial instigation of a natural phenomenon, there is no > instance where a model successfully eliminated the natural physics > underlying it. Indeed the idea itself would be thought very odd. > > Yet in AGI that is what we have been doing for 60 years. It is possible > to see that an exceptionless chain of hundreds/thousands of years > successful technology development (and related science) outcomes now > has, in the modern era, one exception. It is the only instance where we > humans have decided to pretend to create a human technology: AGI. In AGI > design, without exception, we have chosen to use computers and > computing. Indeed such is the pervasiveness of it, if you suggest to AGI > designers that failure will ensue in the same way that computed fire is > not fire, you will elicit incredulity. > When you feel pain from sticking your hand in a fire, is it because the chemical effects of heat have reached your brain, or is it because the nerve signal has been communicated to your brain? What is to prevent part of this signal propagation from being substituted with some other implementation, such as fiber optic cable? How far up into the brain can this swapping be done before it no longer burns? > > "According to the Church-Turing thesis, a Turing machine can emulate any > other Turing machine. The physical Church-Turing thesis claims that > every physically computable function can be computed by a Turing > machine. This is the basis for brain emulation: if brain activity is > regarded as a function that is physically computed by brains, then it > should be possible to compute it on a Turing machine. Even if true, > however, it does not demonstrate that it is a computationally feasible > process." (Sandberg & Bostrom, 2008) > Is it possible in theory or practice to accurately model the behavior of the chips you are developing? If so they are computable, as would be any brain constructed using them. Finding an informational process that cannot be replicated by a computer is a bit like finding a sound that cannot be played by a set of speakers. Speakers are universal sound emitters just as computers are universal processors of information. > > a Turing machine can emulate any other Turing machine..... get it? This > has got ZERO to do with the problem of AGI. Replication, not computing, > will do it... and computing is NOT replication. It is emulation at best. > The key to empirical resolution of all these issues is replication. > Until you play with the real physics ... you are just doing armchair > philosophy and dreaming your answers. > The emulation of the behavior of the cochlear nerves restores the sense of hearing to those that are deaf. If someone had brain damage in their primary auditory cortext, and it was repaired with the aid of a functionally equivalent chip, would you say they are still deaf because it is only simulated hearing, rather than some more true form of hearing? At what point during the substitution of functionally equivalent parts would intelligence no longer be possible? Jason -- You received this message because you are subscribed to the Google Groups "Everything List" group. To post to this group, send email to everything-list@googlegroups.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.
