X-Message-Number: 7924
From:  (Thomas Donaldson)
Subject: reply.previous.cryonet.msgs
Date: Sun, 23 Mar 1997 11:36:45 -0800 (PST)

Hi!

This is my reply to the LAST set of cryonet messages, not to this one. 
Apparently my net node has been fixed. Mike Perry might send a copy of this
same reply, which need not be read.

			Long long life,

				Thomas Donaldson



Hi again!

Here I go again, with a few minutes stolen from my current work on PERIASTRON.

1. Mike, the REASON why digital does not work is not mysterious. You are
   inevitably making a symbolic representation of a person if you use a 
   digital computer. Neither of us is a symbolic representation; I do not
   need your recognition of me to be aware, nor vice versa. 

   It is certainly also true that chaos will cause any such representation
   to behave differently from its original, I suspect in a very short time.
   This is because it cannot really start out with EXACTLY the same parameters
   as the real person. And ALL numbers used in a computer are symbolic
   representations, models of how numbers really behave. 

2. We are not computers and when we normally think we are not computing. True,
   we can imitate computers, do arithmetic, and so on. But the difference is
   essential. I sit here thinking out what I am to say, as a description
   (very distant) of the physical state of my own brain. That physical state
   depends not just on mental entities such as numbers, but on the fact that
   something scratches my thigh, I hear a fan overhead, I feel a bit tired,
   and many other things. To say that we perceive would be more accurate,
   though of course we act on the world and thus change our perceptions.

   Computing of any kind occurs always within a system invented by human
   beings. Long ago we invented numbers, and have gone on from there into
   much complexity. But that system remains symbolic only, and its relation
   to reality comes not from anything intrinsic to it but because we human
   beings find it useful when considering reality. Some day we may even find
   some better way yet unimagined.  

   For a long time there has been a strain of AI which assumed that if we
   can get our computer to handle all the symbols just right, it will become
   intelligent. And the programs such people created could indeed, in very 
   restricted special cases, handle the artificial "reality" presented to
   them. Put them into the real world, and they crashed almost at once:
   too many parameters, too much computation. We are beginning to work our
   way out of that problem with hybrid devices such as neural nets, which
   are not programmed but learn from the stimuli presented to them. As real
   creatures we deal ALWAYS with events which can only be symbolized with
   difficulty... even when we do physics or math. And note the main feature
   of neural nets: they can recognize objects (say handwriting) which would
   be virtually impossible to characterize in any dictionary or with any
   set of symbols.

   (And note that I am not saying that artificial intelligence, or even just
   artificial awareness, is impossible. They are possible. It is just that
   you will no more get them through computation --- no matter how elaborate
   and extensive --- than you can swim to the Moon).

3. As for computers, yes, at best they can "just (clumsily) go through the
   motions". There seems to be an assumption here that the word "(clumsily)"
   which I interpolated should not be there. One of the philosophers on
   this question, discussing the Turing Test, proposed that we do the
   interrogation in Chinese (on a suitable keyboard) and on the other side
   of the wall there would be a person very highly trained from a Chinese
   dictionary --- not one giving English equivalents, just one giving 
   Chinese definitions of Chinese words --- to pick out the proper ideographs
   in answer. That is, this person could play properly with all the symbols,
   but had not the vaguest idea what they might MEAN.

   (Yes, the person would have to be superhuman in a special way --- or
   perhaps a large and very fast computer. WE learn words by figuring out
   what they mean, and see when to use them because we live in the world
   and recognize various real events and their structure. The computer,
   however, does not do it that way at all -- so much the worse for the
   computer, which may have to do billions of machine instructions to work out
   the answer to "Good Day" in Chinese).

   If this computer is aware, I'd like to know just how and why it is aware.
   And even more important, what is it aware of? It certainly isn't aware
   of any conversation in the sense we are aware.

   I think of Mr. Lynch as aware basically because I believe that he has 
   a physical structure and brain structure close enough to mine that he
   will have the same awareness (not of the same things) as I. Perhaps I
   am mistaken. Well, when I was less than 5 years old, I thought that
   the Santa Claus I saw at Christmas time was the real Santa Claus. It
   proves very little to say that I might be mistaken.

Bedtime now. I am sure that none of you will be convinced. But good night.

			And long long life,

				Thomas Donaldson




--AAA19616.859106116/blacklodge.c2.net--

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=7924