X-Message-Number: 12980 From: Date: Tue, 21 Dec 1999 22:32:29 EST Subject: careful--zombies Some brief responses to Daniel Crevier's summary of the arguments for the possibility of consciousness in computers. 1. I agree--and have always said--that consciousness in computers is POSSIBLE, as far as we know. In fact, I don't know of any informed person who categorically denies that it is possible. My argument is with the "strong AI" people who think their case is overwhelmingly strong. 2. >Argument for the reproducibility of conscious behavior in machines. >Let us start with digital computers. Thy can simulate any physical system, and >therefore can simulate a human brain. "Simulation" is a potentially deceptive term. First, there can be partial isomorphisms which one could call simulations, but there would not be one-to-one mapping. Only a small fraction of the numbers generated by the computer correspond to parameters of the system simulated. Second, a digital computer could not (with full accuracy) simulate a continuous universe; and we have no assurance that our universe is discrete at the level of a digital computer. Third, the word "simulation" fudges the time element. If a linear computer "simulates" a dynamic system, then one could equally well say that the mere program and initial data store themselves simulate the system, without the computer running. In fact, this appears to be close to Moravec's position, that "existence" at bottom is just abstract relationships. Fourth (and this is partially redundant), a mere description is not an emulation. I could (in principle, assuming that present quantum mechanics is the be-all and end-all, and assuming enough information and enough time) write out in long hand a description of the present and future states of any system, e.g. you and your environment. To imagine that such a description would BE you and your future life seems like nonsense, apologies to Moravec. 3. >Argument from the uploading thought experiment. >You are operated upon by a robot surgeon which analyses a small part of your >brain, and constructs an electronic circuit (either analog or digital, it doesn't matter) >with the same input-output properties. There is no assurance that this is possible, even in principle. > at the end of the process, all of your brain will have been replaced by circuitry, and >you will have become a 'simulation' of yourself. Yet you will still be conscious. If >not, at what point in the substitution process did you lose your awareness? If consciousness depends on a special organic mechanism (the "self circuit"), you lost consciousness when that was removed or damaged. >Further, if you did lose your awareness, then it slipped away from you without your >being aware or it, because at the end you will still maintain that you are conscious. I don't concede that; see above. In any case, the vocal assertion is not proof. >Losing one's awareness without being aware of it sounds like quite a contradictory proposition. On the contrary--you cannot be aware of losing awareness, by definition. (Well, you might be aware of it fading in some cases, but you could not be aware of having lost it.) 4. >Argument from the unobservability of 'true' consiousness. >If there can never be any observable difference between a zombie and a conscious >being, does it make any objective sense to talk about 'true' vs 'make believe' >consciousness? If it doesn't, acting as if conscious and being conscious must be >one and the same thing. Depends what you mean by "observable." Lots of real differences in systems are not observable in practice. One could also ask about the "reality" of events outside and inside the horizon of a black hole--but our case is simpler. We could, in principle, observe the workings of a "self circuit" with noninvasive scans of one sort or another, or possibly some type of telepathy. Time lurches on, and I'll mercifully stop here, at least for now. Robert Ettinger Cryonics Institute Immortalist Society http://www.cryonics.org Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=12980