Cognitive Science | Seminar: AI: Hope and Hype
Q700 | 1003 | Hofstadter


In this seminar, we—the students and the professor together—will
collectively take a skeptical look at a number of highly-touted AI
projects and overall approaches.  By no means is our purpose to bash
indiscriminately; rather, we wish to give credit where credit is due,
but by the same token to remove credit where it is not due.  Some of
the topics that will come under scrutiny will be:  case-based
reasoning (a number of CBR projects including COACH, MINSTREL, CHEF,
and some others), Rodney Brooks’ robotics project COG, Douglas
Lenat’s CYC project, David Cope’s music-composition program EMI,
Keith Holyoak and Paul Thagard’s analogy-making program ACME, various
computer-driven vehicles projects, and—just for fun—a so-called
“computer-written novel”.  If there is time, we may also turn our
attention to SOAR, some projects in Artificial Life, and a few other
projects.  In addition, we will consider the way AI and AI-related
topics are represented in the contemporary press (for example, press
coverage of the so-called “Turing Test” competition that until
recently, was held annually at the Boston Science Museum, and perhaps
still is being held).
The students will bear a great deal of the responsibility for this
seminar.  In particular, they will be assigned articles or books to
read and to present to the class.  They will have to work hard to
distill the essential ideas from the often-overwhelming flood of
garbled prose and superfluous information.  (This is a skill that we
hope to enhance during the course.) They will also have to listen
critically to each other’s presentations and to ask incisive
questions.  Students’ presentations will be judged for clarity,
pinpointing of essentials, and incisiveness; needless to say, these
qualities will count a great deal toward a student’s final grade.
Students’ questions to other students will also be important factors
in grading.
In addition, a term paper will be required, most likely of about 10
pages in length (of course, the exact page count is not the point),
which takes one AI project or approach and makes explicit its
fundamental implicit worldview (something that is usually not spelled
out or even acknowledged by its authors), sums up its merits and
demerits, and so on, in the clearest possible terms.
The quality and clarity of the written English in the term paper will
play a major role in the student’s grade. After all, a lot of what
this course is about is honest, efficient, and clear communication.
The basic purpose of the course is to try to separate the wheat from
the chaff in this sadly hype-strewn but nonetheless very fertile
field.  To be sure, we will not develop a foolproof decision
procedure for making this subtle distinction, but we hope to gain
insight into the nature of the challenge, so that students can learn
to read (or skim) articles and size up ideas and claims with more
sophistication and confidence than before.  Prerequisites:  At least
one year of coursework in AI or cognitive science.
Preference for admission will be given to graduate students.