Path: utzoo!attcan!uunet!husc6!cmcl2!nrl-cmf!ames!pasteur!ucbvax!decwrl!hplabs!sdcrdcf!trwrb!ucla-an!remsit!stb!michael From: michael@stb.UUCP (Michael) Newsgroups: comp.society.futures Subject: Re: The future of AI Message-ID: <10425@stb.UUCP> Date: 15 Jun 88 19:21:14 GMT References: <48.22A3B84F@isishq.UUCP> <4347@killer.UUCP> Reply-To: michael@stb.UUCP (Michael) Organization: STB BBS, La, Ca, Usa, +1 213 459 7231 Lines: 113 In article <4347@killer.UUCP> elg@killer.UUCP (Eric Green) writes: [Discussion of predicability of humans, and modeling by computers] >To generalize, there are a couple of basic assumptions needed to make >AI a science instead of a religion: > >1) The human mind consists of mechanism (program) and data (memories), >2) All human action is detirmined by the operation of the mind's mechanisms > upon the mind's memories, >3) The computer can model the above. > >In particular, > [ specifics for points 1 and 2] > >> this into an AI model, and ask a machine to decide which is "best" or >> which is "right", the democratic party or the republican party. Try to >> replicate the human decision-making process at the ballot-box. > >Under the above model, your choice is dictated by your past experiences >(memories) and by various mechanisms operating upon those memories (e.g. >mechanism: pleasure/security. Memories: Political discussions, which policies >were best, who was the better speaker, smarter, etc.). > >That is, your choice consists of neurons firing in a particular pattern. And, >if somehow, microseconds before the vote, we had the sum total of your >knowledge and experience available to us, we could predict exactly how you >would vote. You could have voted in no other way besides that dictated to you >by your knowledge and experience and the underlying mechanisms. Careful of what can of worms you open. Allow me to re-phrase you: If we knew exactly what program you operate on, and the data availible to you the moment before you enter the booth, we know exactly what you will do. If we knew exactly what program you operate on, and the data availible an hour ago, and the inputs you received during that hour, we know exactly what you will do. If we knew exactly how the world was to the decimal, and we know your state at that same moment, then we can calculate all the inputs you will receive and determine exactly what you will do at any time in your life. Congradulations! You have just denied the existance of free will. This is a highly debatable point. Determinists hold your position, that everything can be determined exactly. (Quantum mechanics seems to indicate the opposite, that everything is random and statistical at best. Personally, I'm hoping that a third posibility will be discovered that allows free will and freedom of choice.) >> AI wants to build machines that can perform tasks or make decisions as >> well as humans. I think though, that human reason and decision making is >> not mechanical, it is a-mathematical, a-scientific and a-rational. It is >> hinged to something else, human values (variable), passion (wholly >> subjective), emotions (volatile) and sympathies (unpredictable). > >Science asserts the exact opposite: That everything can be explained, that >everything can be modelled, and that the way to gain knowledge is to seek >these explanations and models. Really? We can't even say what an explanation is (i.e., no specific step-A step-B step-C "this is an explanation" algorithim exists that agree with our intuition). Try modeling a single electron sometime--tell me where it will be. >Religion, on the other hand, is completely the opposite. Religion asserts that >there are things that cannot be explained, that some things are beyond >comprehension, and to seek explanation is to lose the faith. Why can't we go faster than the speed of light? "Well, its just a fundamental limit" But why is it a fundamental limit? "It just is" I hope you don't think I'm nit picking here, but... > >To say that artificial intelligence cannot be attained because of magical or >mystical properties of the human mind is thus of no relevance to scientists. >Similiar statements have been made about other scientific endeavors, and have >always proven false. The stumbling blocks to AI are much more mundane: the >large bulk and limited computing power of today's computers, providing enough >input devices to provide a suitable store of memories upon which to operate, >etc. Actually, I have a bigger stumbling block to propose: Turing machines can do anything a major computer can do. There are a countable number of turing machines, but an uncountable number of problems. Therefore, any computer will be limited in what it can do. A computer cannot do things that it is not programmed to do. Humans, however, have hunches, can deduce the existance of a double helix spiral, or a wave-particle combination, without having had it programmed in to them. The point: Beyond a certain point, all you can program a computer to do is try every possible combination til something looks right. Humans have hunches, guesses, flashes of insight, etc. and check only a few ideas before comming close. > >-- > Eric Lee Green {cuae2,ihnp4}!killer!elg > Snail Mail P.O. Box 92191 Lafayette, LA 70509 >"Is a dream a lie if it don't come true, or is it something worse?" Michael p.s. Followup should probably go to a philosophy of science group, but I don't know of any. : --- : Michael Gersten uunet.uu.net!denwa!stb!michael : ihnp4!hermix!ucla-an!denwa!stb!michael : sdcsvax!crash!gryphon!denwa!stb!michael : What would have happened if we had lost World War 2. Well, the west coast : would be owned by Japan, we would all be driving foreign cars, hmm...