00100	AREAS OF DISAGREEMENT WITH LIGHTHILL
00200	
00300	
00400	1.     Lighthill classifies neurophysiology and psychology as at  the
00500	theoretical  pole  opposite  applied  AI.       On  the contrary, the
00600	theoretical side of AI is the study of intellectual  mechanisms  like
00700	reasoning, learning, tree search, modelling, compiling.
00800	
00900	2.  Progress  in  chess  playing  has  not  been  made  by making the
01000	evaluation function more sophisticated.      That  is  what  everyone
01100	expected  at first.  Progress has been made by making the tree search
01200	more sophisticated so that the right positions are investigated.
01300	
01400	3.    Lighthill is right that progress is to be made by putting human
01500	knowledge into programs.  However, this is not easy, especially since
01600	in common sense physics, no-one knows what human knowledge is.  Can i
01700	give an example from common sense hydrodynamics.
01800	
01900	4.  There  is  fairly  universal  agreement  now  that the problem is
02000	representation.     I would emphasize the representation  of  partial
02100	information.
02200	
02300	5. Lighthill has ignored much of the material submitted to him.    In
02400	particular, his classification  of  the  field  ignores  rather  than
02500	disagrees with the contributions of Burstall and Meltzer.
02600	
02700	6.  I have been conservative, and the over-optimism results both from
02800	optimistic personalities and certain ideas of what the problems are.
02900	
03000	7.  The need for robotics is precisely to help avoid confusions  such
03100	as those exhibited by Lighthill.