Path: utzoo!attcan!uunet!know!zaphod.mps.ohio-state.edu!mips!dimacs.rutgers.edu!aramis.rutgers.edu!athos.rutgers.edu!christian From: starpath@athena.mit.edu Newsgroups: soc.religion.christian Subject: AI Message-ID: Date: 13 Nov 90 09:27:29 GMT Sender: hedrick@athos.rutgers.edu Lines: 86 Approved: christian@aramis.rutgers.edu > __ > /..\ In quest of knowledge.... > --mm--mm-- Mike Gobbi Well, good. Here is a bit for you: >No matter how sophisticated programs and computers get, they will never be >conscious as we understand the term (I am a computer science student and have >studied this question in on of my courses, so I am pretty confident in my >statement). The programs will no more have "morality" than does an animal >trap. Now that is a pretty strong statement. In fact, it is pretty mistaken. I really am not familiar with Rutgers CS program, but I am somewhat familiar with MIT's CS program, it includes/has included Marvin Minsky and Seymour Papert, two rather big names in AI. (No, I'm not CS, I'm EE, so I'll have to paraphrase.) Minsky believes that while the AI approaches so far do fairly well at answering specific problems, (IE expert systems) there one major problem: such systems cannot 'learn' in the sense of creating new abstractions based on old information. Of course, current systems can create new things, but only by following some pre-programmed set of algorithms. His hope is to eventually build a system complicated enough to be able to learn much like a child can; that is, give the system only one axiom: the ability to learn "things," instead of some specific list of things. Only this system would have the ability to truly learn, and therefore think. Now Minsky thinks this is possible. Papert, who is also a great man in the field of early childhood development, thinks that this is possible. No one has been able to refute them. Lots of philosophers have come up with thought-experiments to show that the intellect cannot be contained only within our skulls. Unfortunately, none of these have held. While I must admit I do not know such things by name, if you send me information as to why you think such a machine will never exist, I can probably put together enough to show you what errors there are in the argument. >Some PEOPLE will write programs that come to "immoral" conclusions, and others >will write software that comes to "moral" conclusions. Nobody teaches the >computer what is right or wrong -- that is built in by the designer and the >users. The problem is that systems that are complicated enough to have AI cannot be 'programmed' in our normal understanding of the word. It would be roughly equivalent to the human non-ability to have control over heartbeat rate, digestion system, etc. An AI computer cannot be "written" to do something 'immoral' by quickly changing the code. Intellegent systems are only partially genetic; the environement plays a very large role in development. > I suspect that the decisions these computers make will be EXACTLY the same >decisions that humans in the same situation would make (only faster). Thus, >if you want to enforce morality, you have to ensure that the laws are moral. Cute, but not true for real AI. This would be true for an expert system, one which takes data only from it's creator. An AI computer is not so intrinsically linked to it's creator. > Coincidentally enough, this is already being done. The abortion issue and >euthenasia issue are two medical problems that spring to mind. How can we >complain that a computer has issues an immoral judgement when we allow a >person to issue the same judgement now? On the legal side there are many >questions relating to homeless, jobless, and opressed situations. Just as in >the medical profession, there is no consensus here on what is correct. But the computers you mention here are not making 'decisions.' The answers the spew are not powerful enough to call them that; it is merely the end result of an algorithm. A 'judgement' is weighing the issues and making a decision based on the issues. The judgement is made when the code was written, not when the answer is processed. An AI machine would really make a judgement. > I think worrying about what computers MIGHT do is far less important than >worrying about what society IS doing right now. That's for sure! Well, thank you for listening to me. I'm sorry I have to go now...my human user wants to log onto me. From the account of: David Hollingsworth starpath@athena.mit.edu [Just for reference, Mike Gobbi is at UBC, which I take it is University of British Columbia, not Rutgers. --clh]