The \F1Lighthill Report\F0 is organized around a classification
of AI research into three categories:

	Category A is \F1advanced automation\F0 or \F1applications\F0,
and he approves of it in principle.  Included in A are
some activities that are obviously applied but also activities like
computer chess playing that are often done not for themselves but
in order to study the structure of intelligent behavior.

	Category C comprises studies of the \F1central nervous system\F0
including computer modeling in support of both neurophysiology and
psychology.

	Category B is defined as "building robots" and "bridge" between
the other two categories.  Lighthill defines a robot as a program or
device built neither to serve a useful purpose nor to study the central
nervous system which obviously would exclude Unimates, etc. which are
generally referred to as industrial robots.  Emphasizing the bridge aspect
of the definition, Lighthill states as obvious that work in category B is
worthwhile only in so far as it contributes to the other categories.

	If we take this categorization seriously, then most AI researchers
lose intellectual contact with Lighthill immediately, because his three
categories have no place for what is or should be our main scientific
activity - \F2studying the structure of information and the structure of
problem solving processes independently of applications and independently
of its realization in animals or humans\F0.  This study is based on the
following ideas:

	1. Intellectual activity takes place in a world that has a certain
physical and intellectual structure: Physical objects exist, move about,
are created and destroyed.  Actions that may be performed have effects that
are partially known.  Entities with goals have available to them certain
information about this world.  Some of this information may be built in,
and some arises from observation, from communication, from reasoning, and
by more or less complex processes of retrieval from information bases.
Much of this structure is common to the intellectual position of animals
people and machines which we may design, e.g. the effects of physical actions
on material objects and also the information that may be obtained about
these objects by vision.
The general structure of the intellectual world is far from understood, and
it is often quite difficult to decide how to represent effectively the information
available about a quite limited domain of action even when we are quite
willing to treat a particular problem in an \F1ad hoc\F0 way.

	2. The processes of problem solving depend on the class of problems
being solved more than on the solver.  Thus playing chess seems to require
look-ahead whether the apparatus is made of neurons or transistors.
Isolation of the information relevant to a problem from the totality
of previous experience is required whether the solver is man or machine,
and so is the ability to divide a problem into weakly connected subproblems
that can be thought about separately before the results are combined.

	3. Experiment is useful in determining what representations of
information and what problem solving processes are needed to solve a
given class of problems.  We can illustrate this point by an example from
the \F1Lighthill Report\F0 which asserts (p. 15) that the heuristics of a chess
program are embodied in the evaluation function.  This is plausible
and was assumed by the first writers of chess programs.
Experiment showed, however, that the procedures that select what part of the
move tree is examined are even more important, i.e. when the program errs
it is usually because it didn't examine a line of play rather than because
it mis-evaluated a final position.  Modern chess programs concentrate on this
and often have simpler evaluators than the earlier programs.

	4. The experimental domain should be chosen to test the adequacy
of representations of information and of problem solving mechanisms.  Thus
chess has contributed much to the study of tree search; one Soviet computer
scientist refers to chess as the \F1Drosophila\F0 of artificial intelligence.
I think there is much more to be learned from chess, because master level
play will require more than just improving the present methods of searching
trees.  Namely, it will require the ability to identify, represent, and
recognize the patterns of position and play that correspond to "chess ideas",
the ability to solve some abstractions of positions (e.g. how to make use
of a passed pawn and a seventh rank rook jointly) and to apply the result
to actual positions.  It will probably also require the ability to analyze
a problem into subproblems and combine the separate results.  (This ability
is certainly required for a successful \F1Go\F0 program).