for ta:
get up ekl maybe
see to enrollment
get handout fixed up and printed and distributed


Notes for first lecture

There are several approaches to AI, and the one I shall cover in
this course is at one end of one  spectrum.  The other end of
the spectrum is to discover enough about human neurophysiology
to make system that imitates the nervous system.  Another
approach is to discover how human intelligence works by
psychological experiments, brilliantly conceived and executed.
The present approach involves discovering enough about the
common sense world to make programs that can achieve goals
and exhibit other intelligent behavior.  From this point of
view, AI is a branch of computer science rather than a
branch of physiology or psychology.

We can even consider AI as analogous to linear programming.
Indeed if the world presented problems to us in the form
of minimizing a linear function of some variables subject
to linear inequalities as constraints, then AI would
coincide with linear programming.  Indeed if there were
a systematic procedure for reducing deciding what to
to in the common sense world to linear programming
problems, then AI would still reduce to linear programming.

Unfortunately, there is still no known mathematical
domain of problems to which there is a known systematic
method of reducing problems of achieving goals
in the common sense world.  We believe that the problem
of finding objects characterized by formulas of
first order logic may turn out to be such a domain
because of its possessing some kind of universality,
but we are far from being able to prove it, not even
being able to express in a definite form what would
have to be proved.

1987 Oct 15

Topics I would like to cover

1. Formalization of heuristics

2. Return to  can.

second batch of papers
Moore
Kowalski
Dennett
Halpern