From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!batcomputer!cornell!rochester!yamauchi Mon Dec  9 10:48:28 EST 1991
Article 1907 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!batcomputer!cornell!rochester!yamauchi
>From: yamauchi@cs.rochester.edu (Brian Yamauchi)
Subject: Clearing Up the Syntax/Semantics Confusion
Message-ID: <YAMAUCHI.91Dec6044536@heron.cs.rochester.edu>
Sender: yamauchi@cs.rochester.edu (Brian Yamauchi)
Nntp-Posting-Host: heron.cs.rochester.edu
Organization: University of Rochester
Date: 6 Dec 91 04:45:36

It occurs to me that the reason why the pro- and anti-Searle debators
are talking past one another is because each side is thinking about a
different type of intelligent system.  Members of the pro-Searle camp
seem to be thinking about traditional symbolic reasoning systems,
while members of the anti-Searle (pro-Strong AI) camp are considering
intelligent machines in general.

An example may make this more clear:

Consider a traditional logic-based reasoning system containing
information about automobiles.  This could include a statement like
the following:

drives(x,y) -> driver(x) ^ car(y) ^
  (exists z: road(z) ^ drives-on(y,z)) ^
  (exists a,b: place(a) ^ place(b) ^ drives-from(x,a) ^ drives-to(x,b))

Searlists would argue that this system doesn't really understand
anything about cars, because one could just as easily replace the
predicate names in the following fashion and represent completely
different information:

paints(x,y) -> artist(x) ^ painting(y) ^
  (exists z: subject(z) ^ represents(y,z)) ^
  (exists a,b: critic(a) ^ critic(b) ^ is-loved-by(x,a) ^ is-hated-by(x,b))

Now, instead of a system that reasons about cars, consider a system
that *drives* a car.  This system could make decisions based upon
perception (e.g. obstacle detection), internal knowledge (e.g. map
information), and combinations of both (e.g. landmark recognition).

In this case, it doesn't matter what the symbols used by the program
are called.  The semantics do not result from the interpretation of
symbol names, but rather from the interaction of the system with its
environment.

You can change the names of the variables used by the NAVLAB all you
want, but you still won't get it to paint the Sistine Chapel...


