Re: blow away?

phayes@cs.uiuc.edu
Message-id: <199309141905.AA04418@dante.cs.uiuc.edu>
Sender: phayes@dante.cs.uiuc.edu
Date: Tue, 14 Sep 1993 14:08:23 +0000
To: interlingua@ISI.EDU, schwartz@iota.cs.fsu.edu,
        sowa <sowa@turing.pacss.binghamton.edu>
From: phayes@cs.uiuc.edu
Subject: Re:  blow away?
John Sowa writes-

>....  The only thing that any of us are proposing is a version
>of first-order logic with a LISP-like and a graph-like notation.
>If anyone else prefers some other notation, they're welcome to
>design it, as long as it maps to the same semantic base.

The trouble with this is that we know pretty clearly that first-order 
logic is inadequate as a semantic base. We need more expressive (or 
perhaps just differently expressive) languages to do a good job of 
knowledge representation. I bet we need some higher-order 
expressibility, and we certainly need some kinds of nonmonotonicity. 
AI - thats the actual science of AI - needs to be developing and 
experimenting with all kinds of representational ideas. It makes 
no sense at all to be talking of 'standardising' a semantic base 
right now: its like physicists in the eighteenth century deciding 
to standardise the atomic table.

>As far as research goes, people are free to do anything they like.
>But none of us are going to be free to do any research unless there
>is enough money coming in to pay the rent.  And money requires
>applications.  And applications require standards.

This is too simplistic. Applications of AI do not always require 
standards. What requires standards is making Kbases transparent to 
communication. This is of course well worthwhile as a goal,and 
probably needs to be done. But it is not a necessary requirement in
order that the whole field of AI should not waste away.

Incidentally, I think we need to be a bit careful. AI in the past 
has been a little careless in its predictions of success, and the 
present tightness in funding reported by many is partly a reaction 
to that overenthusiasm. We should not tell anyone, or even let anyone 
believe, that we know how to standardise Krep because we have all the 
problems of knowledge representation solved. 

Pat Hayes


----------------------------------------------------------------------------
Beckman Institute                                    (217)244 1616 office
405 North Mathews Avenue        	   (217)328 3947 or (415)855 9043 home
Urbana, IL. 61801                                    (217)244 8371 fax  	  

hayes@cs.stanford.edu  or Phayes@cs.uiuc.edu