00100	I would really like to know what is Leibniz argument; could you  make
00200	a  summary,  Yorick?   There seems to be an implicit argument that AI
00300	could not mean anything to philosophy, not that that bothers me.  But
00400	philosophy  is said to be concerned only with non-contingent matters.
00500	If that particular view of philosophy is interesting, then  it  gives
00600	up  vast  areas which traditionally were philosophy and which are now
00700	physics, psychology, physiology.  Virtually  anything  I  care  about
00800	seems  to  be  contingent, and particularly what we are capable (in a
00900	hardware sense) of seeing, perceiving and representing.  If  none  of
01000	this  experimental  epistemology is relevant to philosophy, then what
01100	is non-contingent?  It would seem that only formal systems.   In  our
01200	systems  we  don't  capture the world, only our model of it.   But in
01300	some way, philosophy is not allowed to ask whether the model  behaves
01400	like  the  world  out  there, only like any possible world out there.
01500	How do we choose axiom systems?  I realize that there is a lot to  do
01600	with formal systems, but that field is crowded, what with mathematics
01700	and AI both involved.   I  would  like  to  talk  about  a  practical
01800	epistemology  for an intelligent being. Richard points out that if we
01900	could show that the human brain can be imitated  by  a  finite  state
02000	machine,  that  tells a lot.  He also points out that the presumption
02100	behind AI (and any science) is that an enormous part of the  universe
02200	can be modelled by some formal system.  But does philosophy not allow
02300	itself to care about whether the formal  systems  model  this  world,
02400	these humans?
02500	
02600	
02700	10 I first don't know what the proceduralist position is.  If we  are
02800	talking  about  work  done  at  MIT,  I doubt that there is a clearly
02900	defined position.  What appeared to me while I  was  there  was  that
03000	there  was  a  real  concern about predicate calculus formulations of
03100	knowledge, that the language was inadequate, that the information was
03200	mixed  up  in  a sea of undifferentiated statements, and that program
03300	control was necessary.  There was no  sense  of  denying  declarative
03400	information,  only  that  adding procedures was a powerful and simple
03500	way of adding control while localizing the  context  of  information.
03600	There was a zeal, also, which is not unique.