From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!uwm.edu!linac!mp.cs.niu.edu!rickert Thu Apr 16 11:33:42 EDT 1992
Article 5013 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!uwm.edu!linac!mp.cs.niu.edu!rickert
>From: rickert@mp.cs.niu.edu (Neil Rickert)
Subject: Re: syntax and semantics
Message-ID: <1992Apr9.181334.4948@mp.cs.niu.edu>
Organization: Northern Illinois University
References: <92099.200726JPE1@psuvm.psu.edu> <1992Apr9.005331.23376@mp.cs.niu.edu> <92100.121650JPE1@psuvm.psu.edu>
Date: Thu, 9 Apr 1992 18:13:34 GMT
Lines: 67

In article <92100.121650JPE1@psuvm.psu.edu> <JPE1@psuvm.psu.edu> writes:
>..
>     I never stated that my approach was designed to "deny the possibility
>that computers could ever have semantics".  I simply wanted to distinguish
>between machine (or biological) level causal interactions and meaning.  The
>distinction only denies the possibility of semantics for both humans and
>computers if one assumes that, whatever meaning is, it is reducible to
>mechanical functions.  I happen to think that its not, and I take my guidance

  Ultimately all these discussions of whether a computer
		can or cannot have semantics
		can or cannot understand
etc, become pointless.  I note your use of the words "whatever meaning is"
and this is at the heart of the problem.

 Neither you nor I know what "meaning" is, what "semantics" are, what
constitutes "understanding".  We both have somewhat hazy ideas, and there
may be a great deal of similarity between our ideas.  But we cannot pin
these down to the point of giving a precise definition.

 Without knowing precisely what is "meaning" the question "can a computer have
meaning" has little real content.

 If you look at AI the right way, it is guaranteed to be at least a partial
success.  For the effort at producing AI will bring us closer to knowing
what really is involved in semantics.  And this will happen even if the
ultimate problem of AI is never solved.  Already from attempts at AI we
know that some ideas about semantics are naive and don't work out when
you try to program them.

>     Please note that in my more extended response to your other comments,
>I stated that my arguments had "not, of course, show[n] that our brains
>'have meaning' in some other way than the computer's state does".  I am
>not at this point entirely convinced wither way as to whether a computer
>can be created and programmed to instantiate the kind of subjectivity

 It does seem to me that to be "not entirely convinced either way" is currently
the only justifiable position.

>found in humans.  I do, however, think it is useful to distinguish between
>formal systems (or computers) and their interpretations.

 Most of my reactions to your comments are on your insistence on identifying
a computer with a formal system.  I would much rather view it as a general
purpose mechanical tool for simulation.  It happens that one application of
that mechanical tool is to simulate a formal system.  But you can better
understand computers if you also recognize they are tools which have many
other uses (such as printing mailing labels on all that junk mail you
receive).  Once you connect a computer to real world inputs and outputs,
you can no longer view it as only a formal system.

 Can a formal system have meaning?  I'm quite willing to accept that it
cannot.  Indeed the whole point of a formal system is that it have no
inherent meaning so that it can be employed in diverse ways.  But that
nothing to do with whether a computer connected to real world inputs and
outputs can have meaning.

 The question "can a computer have meaning?" is just the wrong question.
It is in the attempt (whether successful or not) to implement
meaning/understanding/etc that we will gain a better idea of what actually
constitutes meaning and understanding.

-- 
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
  Neil W. Rickert, Computer Science               <rickert@cs.niu.edu>
  Northern Illinois Univ.
  DeKalb, IL 60115                                   +1-815-753-6940


