In message <001801be35b2$bc6c0840$e2194018@CS1000568-A.cgno1.ab.wave.hom
e.com>, David McFadzean <morpheus@lucifer.com> writes
>
>-----Original Message-----
>From: Robin Faichney <robin@faichney.demon.co.uk>
>To: virus@lucifer.com <virus@lucifer.com>
>Date: Friday, January 01, 1999 3:05 AM
>Subject: Re: virus: kurzweil cuts the mustard
>
>>information currently arriving via my modem. A machine
>>could do *anything* without any consciousness whatsoever.
>
>A machine could not truthfully announce "I am conscious"
>without being conscious. Or is that cheating? :-)
Well, it is, actually, because it boils down to saying the machine couldn't be conscious without being conscious. :-)
>Let's say the machine in question could successfully keep
>up its end of the conversation about what it is like to
>have consciousness, i.e., it passes the Turing test. Sure,
>it is possible that it has a number of preprogrammed
>answers to all my questions, but given the number of
>possible questions it would seem astronomically unlikely
>that it lucked out and passed the test by cheating.
Nope. There's no reason it has to be a lookup table. Any kind of symbolic reasoning and/or neural nets is allowed. Passing the Turing Test does *not* imply consciousness. All it means is, it would be natural for us to think of it as conscious -- which tells us something about us, but as for the machine -- it has to be intelligent, yes, but conscious, no.
>So saying that a machine could do *anything* without
>consciousness it rather like making the claim that it
>is possible for a human to walk through a solid wall
>(you just have to luck out and line up your atoms
>correctly).
I should have said "anything within reason". Obviously, logical and physical impossibilities are ruled out.
-- Robin