utility@quiche.cs.mcgill.ca (Ronald BODKIN) (01/11/90)
(More on the Chinese Room) One thing I'm curious about is why everyone ignores the book with the instructions in it when it comes to understanding. If I have a computer that "understands" Chinese, with a cpu and a memory and I sever the two there is no way that either would understand. Likewise, Searle provides the processor and the book provides the memory. The system understands, and if it seems weird to have understanding disembodied in a system like this, then its also weird to having processing severed. If Searle could MEMORIZE these rules, it probably still wouldn't be enough, because he is used to understanding in a very different manner of operation (i.e. its hard to make a plant grow like a frog does, but the both of them grow) -- and moreover his understanding of Chinese is entirely isolated from the rcontext of his usual knowledge, so it doesn't seem like understanding at all to a more advanced creature (then again, compared to a university math student, one can argue convincingly that a grade 2 student learning multiplication doesn't "understand" multiplication and more importantly IF the math student at university had that little understanding he wouldn't really consider himself to understand at all). So in some sense, the claim that Searle doesn't understand is partly a manifestation of our OWN inability to imagine a person operating like a computer and understanding. As for the point about computers which fake understanding so well that they pass a Turing Test, my question is why does it even matter if they don't "understand" but they can respond appropriately in every case. Or how about a computer that is so intelligent it ALWAYS wins (i.e. fools the examiner) -- even if he tries to use "reverse- psychology" on it? Such a machine might start talking about us only simulating computers but I'm straying from my point. And essentially, that is to say that if aa book had the instructions for a process which enabled people to act intelligently enough to pass/win a turing test and it could also give a prescription for how to save a man's life which wasn't otherwise available, I'd be willing to bet that people would be happy to execute that process and that the process (or at least many processes complex enough to be interesting) cannot be executed correctly without an understanding. Ron
muttiah@cs.purdue.EDU (Ranjan Samuel Muttiah) (01/11/90)
In article <1953@quiche.cs.mcgill.ca> utility@quiche.cs.mcgill.ca (Ronald BODKIN) writes: >(More on the Chinese Room) > One thing I'm curious about is why everyone ignores the book >with the instructions in it when it comes to understanding. If I have >a computer that "understands" Chinese, with a cpu and a memory and I >sever the two there is no way that either would understand. Likewise, >Searle provides the processor and the book provides the memory. The >system understands, and if it seems weird to have understanding disembodied >in a system like this, then its also weird to having processing severed. For a more provocative reading on this checked out: Mind and Brain, the many-facted problems Ed. Sir John Eccles. - Pay close attention to the chapter by one J. Pringle(Ox. Zoologist) and the commentary by J. Josephson (Cam. Physicist. Yes, the Josephson junction man).
kmcentee@Apple.COM (Kevin McEntee) (01/12/90)
In article <1953@quiche.cs.mcgill.ca> utility@quiche.cs.mcgill.ca (Ronald BODKIN) writes: >(More on the Chinese Room) >So in some sense, the claim >that Searle doesn't understand is partly a manifestation of our OWN >inability to imagine a person operating like a computer and understanding. > Ron I have to agree here. I only hope that AI research does not limit itself by only recognizing intelligent artifacts that are analogous to presently accepted biological brains. This criterion of intelligence, by analogic behavior, might close our eyes to a radical discovery of machine intelligence. Kevin kmcentee@apple.com