From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!think.com!yale.edu!jvnc.net!darwin.sura.net!europa.asd.contel.com!uunet!mcsun!uknet!edcastle!aiai!jeff Tue Mar 24 09:56:56 EST 1992
Article 4565 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!think.com!yale.edu!jvnc.net!darwin.sura.net!europa.asd.contel.com!uunet!mcsun!uknet!edcastle!aiai!jeff
>From: jeff@aiai.ed.ac.uk (Jeff Dalton)
Newsgroups: comp.ai.philosophy
Subject: Re: The Systems Reply I
Message-ID: <6431@skye.ed.ac.uk>
Date: 18 Mar 92 18:10:28 GMT
References: <1992Mar12.001918.2564@ccu.umanitoba.ca> <6423@skye.ed.ac.uk> <1992Mar18.072634.9259@ccu.umanitoba.ca>
Sender: news@aiai.ed.ac.uk
Organization: AIAI, University of Edinburgh, Scotland
Lines: 84

In article <1992Mar18.072634.9259@ccu.umanitoba.ca> zirdum@ccu.umanitoba.ca (Antun Zirdum) writes:
>In article <6423@skye.ed.ac.uk> jeff@aiai.ed.ac.uk (Jeff Dalton) writes:
>>In article <1992Mar12.001918.2564@ccu.umanitoba.ca> zirdum@ccu.umanitoba.ca (Antun Zirdum) writes:
>>>In relation to the above, what would the ANTI AI crowd require
>>>so that a symbol has a reference (semantics). If you do not
>>>know what it means to have semantics, then how is it possible
>>>for you to argue that something does not have it?
>>
>>You might start by showing how "cats" lines up with cats and
>>not with cherries.  For which see the Putnam discussion now
>>lost in the noise and perhaps abandoned.

Decided to skip this part, eh?

>>It certainly requires something more than behavior.  To anyone
>>who thinks that behavior is all that is required, I don't think
>>there's anything more to say.
>>
>Perhaps we can agree on this point, but I really need to know
>what is that *thing* that is required for understanding?????

That is, unless I can tell you how human understanding works
in detail, machines must have it too?  I'm sorry, but it just
doesn't follow.

(BTW, it should be obvious that more than behavior is involved.
Do you just behave, or do you have thoughts too?)

>>We know humans can do it.  And there are arguments that machines
>>can't (just by running the right program).  The correctness of
>>those arguments does not depend on knowing how humans do it.
>>
>All of those arguments boil down to this "machines can't because
>they are not people!"

I'm sorry that you can't see anything more in them than that.

If you boil them down, what you ought to come up with is that
since people can understand and machines can't, people are not
machines.  (Insert usual qualifications about understand merely
by running the right program.)

Quite simply, the arguments are _not_: machines are not people,
therefore they do not understand.

Here's a similar case.  Suppose there's an argument that a particular
kind of airship is too heavy (or wrongly shaped) to fly.  This is
_not_ an argument that it is not a bird and hence it can't fly.

>What if I showed you that people are machines?

Depends on just what you showed.  But look at Searle's remarks
on "meat machines" first.  Indeed, I recommend (again) the first
Reith Lecture, reprinted as chapter 1 of his _Minds, Machines,
and Programs_ (or some similar title).

>>>Take the word 'BLUE' (color) what does it mean to say that you know
>>>what 'blue' means? Does this mean that you have something in your
>>>brain that is blue,
>>
>>Are you serious?
>>
>Dead serious!

Really?  You think the "something blue in your head" is a reasonable
suggestion?

>Now please answer the question, how does one understand blue.

That is, unless I can tell you how human understanding works
in detail, machines must have it too?  I'm sorry, but it just
doesn't follow.

>Please! It is necessary to know what understanding is
>to argue about it!

To the extent that I want to do this, I have already done it in
the "definition of understanding" thread.

So I will leave it to you to present the definitions.  If you
think this is unsatisfactory, I suppose that's understandable,
but I don't think there's any hope of us agreeing in any case.

-- jd


