Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!cs.utexas.edu!romp!auschs!d75!awdprime!piobe.austin.ibm.com!sjb
From: sjb@piobe.austin.ibm.com (Scott J Brickner)
Newsgroups: comp.ai.philosophy
Subject: RE: UNIFIED MODEL FOR KNOWLEDGE REPRESENTATION? (IMPOSSIBLE
Message-ID: <8455@awdprime.UUCP>
Date: 13 Jun 91 20:17:05 GMT
References: <25348@samsung.samsung.com>
Sender: news@awdprime.UUCP
Reply-To: sjb@piobe.austin.ibm.com
Organization: IBM Austin, Contractor
Lines: 35


I'm not anything like an expert in AI, or even philosophy, but as an
independent thinker, I've got to poke my nose in here for a second.  The
thread which got us here had some argument about the value of a
knowledge-base that was based only on the subject's "perception" of the
world around him... the question proposed was "How, then, can any two
speakers know that they are speaking about the same thing?"

How, indeed?  This is a VERY fundamental philosophical question... How
do you know ANYTHING?  What does it MEAN to KNOW something?  I think
that thinking too much in terms of an AI knowledge-base leads us
sometimes to forget that they are supposed to be (in some sense) similar
to "natural" intelligences (NI), with our own intelligence being the
only model of such available.  So the first question then, is "What does
it mean to know something?"  One proposed answer is that the ONLY real
facts you know about the universe-of-discourse (universe) is what you
"perceive".  The implication here is that the NI's knowledge-base is
purely his PERCEPTION of the universe.  A corrollary is that it is the
ACT of PERCEIVING that CREATES THE REALITY within the universe.

Two such subjects only know about objects about which they have had
common perception. It is to be hoped that the universe is sufficiently
consistent that there is normally some non-empty intersection between
these two perception sets.  Each intelligence also perceives objects
indirectly, through communication with other intelligences.  Eventually,
a process of abstraction creates an object-image within the knowledge
base of "what everyone knows about such and such".  THIS is how we know
we're talking about the same thing (or at least THINK we are).

So what's the problem?

Sorry if this is somewhat elementary, but it seems relevant to me (i.e.
I perceive it as relevant).

Scott