### ### ### ### ### #### ### ### ### #### ### ### ##### ### ### ### ### ### ### ### ### ### ##### ### ### ########## ### ### ########## ### ### ### ### Underground eXperts United Presents... ####### ## ## ####### # # ####### ####### ####### ## ## ## ## ##### # ## ## ## ## #### ## ## #### # # #### ####### ####### ## ## ## ## ##### # ## ## ## ## ## ####### ####### # # ####### ####### ####### [ It Is Not Obvious That A Machine Can Think ] [ By The GNN ] ____________________________________________________________________ ____________________________________________________________________ IT IS NOT OBVIOUS THAT A MACHINE CAN THINK by THE GNN/DualCrew-Shining/uXu It is popular for AI-researchers to claim that a mere computer program can simulate human thinking. The argument is quite easy to understand: since the mind is nothing more than the human brain, and the human brain is nothing more than cells, and cells are nothing more... etc., we need only copy the data of the smallest parts into a computer program and thus gain a fully-working model of the human brain. If we conceptually connect 'intelligence' with 'thinking' there is no problem. A pocket calculator 'thinks' within this definition. However, if we connect 'thinking' with 'consciousness', we enter a different realm. A pocket calculator has no consciousness; it has no concept about its own thinking. Can we gain the same sort of consciousness that we will find in a human being by merely copying the data of the brain into a computer program? Yes, some say - and repeat: since the consciousness is built upon the construction of the brain, we need only to copy the smallest parts into the program and so-and-so... we will get something that is identical to the human mind. But unfortunately, this is not a fact. If we want a consciousness that is like the human brain, a computer program will not do. The reason why the scientists believe this is because there is an inherent ambiguity in the very argument about identity that they use. To see this, let us distinguish between two different kinds of identity: A-identity: The color of yellow requires the chemical constitution ABC. I.e., if we mix A, B and C we get a yellow paint. However, we can obtain the same kind of yellow color by mixing G, N, and Q. ABC and GNQ are two different sorts of chemical constructions but gives us the same kind of yellow. They are A-identical. B-identity: Say that I enter a machine that scans my physical constitution, and constructs a copy of me with the help of the same sorts of particles. The person that I would meet after the copying would be exactly like me. We would be B-identical, our chemical particles would be of the same sort, but it would of course not be the same particles (I mean, I use them myself!). It is obvious that A-identity and B-identity are not of the same kind. It would be wrong to claim that ABC and GNQ are the same sort of chemical constitution: it is a contradiction in terms. But me and my copy are of the same chemical constitution. We _are_ of the same kind, contrary to the two yellow colors that merely _look_ like as if they were of the same kind. When one copies 'the data of a human brain into a computer program' one gains A-identity, not B-identity. Because a computer program is not like the human brain in its physical constitution, which is a fact. However, the AI-scientists seems to believe that A-identity and B-identity is virtually the same thing. They mix their very way of gaining A-identity with the way one gains B-identity. All properties that belong to me as a person are in the copy of me if and only if we are B-identical. If we are merely A-identical, this is not the case. Now, some might say that the whole problem is trivial. Of course, the computer program will not be exactly like a human mind. It is, after all, not a brain but a program! But has the program a special sort of consciousness that need not be like the human mind, but alike? This is not obviously so (but then, it is not impossible either). Only because we have an A-identity between x and y, whereas x has the property p, it does not automatically (or conceptually) follow that y has, by necessity, this property too. Therefore, it is not correct to suppose that a mere computer program that simulates the human brain has consciousness. Something that would follow quite naturally from the above is a clear definition of what a 'consciousness' is - but that I do not have. But on the other hand, that is a completely different question. --------------------------------------------------------------------------- uXu #392 Underground eXperts United 1997 uXu #392 Call THE YOUNG GODS -> +351-1XX-XXXXX ---------------------------------------------------------------------------