Student Impressions
This page collects impressions from the student
questionnaire surveys. We show all comments from the students, good and bad.
Schools covered are
American University,
Berkeley,
Buffalo,
Columbia,
Harvard,
Oregon State,
Pittsburgh,
and the collected online responses from various sources.
American University Student Impressions
Very interesting book: most of the book was very readable.
Chp 9 needs improvement - more solved examples would help.
Book is useful, but Chps 26 & 27 are too hard.
Very high level and anlytical - entices the mind to think
abstractly
Good graphical inserts would help in explaining key points.
The book is useful, but not easy to read or understand.
the book is useful, good format, and fairly easy to read
and understand.
The book is interesting, useful, and easy to read. Some
topics are well handled.
very useful.
book is very useful and moderately easy.
The book is very good and helpful. I enjoyed reading most
of it.
Good mix of theory and applications.
U.C. Berkeley Student Impressions
Does a good job of teaching AI...except the algorithms for
intractable problems, it doesn't seem very applicable to
other CS fields.
Way too much jargon. I was drowning in information that
is irrelevant for the level at which I'm trying to understand
AI. If you put all that info in little side boxes and label it
as irrelevant to the basic understanding of what I need to know,
then I can deal with it. Just don't clutter it around the info
I need to pass a midterm.
excellent!
Good. Very readable and informative.
Very positive impression. It nicely puts together concepts that would
seem unrelated to each otherwise.
Well written; the humor was great... if you can add more ancedotes or funny
examples they really work as good things to remember.
pretty good
I think it is very helpful in understanding the concepts of the class. There
is a lot of material for the professor to cover, and it can't all be done in
lecture. The book does a good job of bridging the gaps in most cases.
I think this is a great textbook for the class... I liked the way the
writing style a lot... it showed a sense of humor and it was a great way to
present the material.
The photos and diagrams were the worst parts of the book. The artwork was
unappealing (my kid brother could draw a better vaccuum cleaning
agent--heck, the authors should be able to make an illustrating agent who
could do better) and looked very un-professional--maybe this is not
important from a technical point of view, but it was a definite turn-off
to me. Most of the writing was easy to follow
but the personal nature made me feel a little uncomfortable.
Looking down to a footnote that says something like "One of the authors is
still impressed when his car beeps at him to pull his keys out of the
ignition" was something I had never encountered before. I'm not sure if I
liked it or not. The book also had a variety of humorous examples which
were cute and a good laugh once in a while. The greatest strength of the
text, in my opinion, was its coverage of such a wide variety of topics.
The accompanying software was no fun to use.
The grid environment package should
have been implemented graphically--a much more natural approach in my
opinion, and the software should have come with more documentation. Having
the wumpus world implemented graphically with color and/or some simple
animations and bitmappend characters would have made dealing with the
programming much more fun--turning on the display option and watching the
grid print out one character at a time is absolutely ridiculous.
My major concern with the book was that there were few coded examples to
look at to see how some of the strategies might be implemented. In
particular, I found that the idea of a DFS using less memory that a BFS
was subtle and a few friends of mine and I did not understand it until
late in the course (of course, it was easy to recite on the test how much
memory and time each used without really knowing why). We finally
decided, maybe even incorrectly, that the algorithms in question must be
generating nodes on the fly--if the nodes in a search problem pre-existed
and resided in memory, it would seem that the difference between a DFS and
BFS in terms of memory requirements would be insignificant. With a coded
example. this might have been easier to see. Even a pseudo-coded example
would have been nice (if the pseudo code were sufficiently detailed).
Having examples of these search algorithms implemented graphically with
simple animations might also be conducive to learning, but the effort is
probably better spent elsewhere. Strangely enough, in CS164 this semester we
used a compiler chapter from another AI book of Norvig's. This chapter
had a large amount of code (definitely a lot of compiler code considering
it was an AI book) and was light on the intelligence side of things. I am
not sure why the authors decided to take such a different approach in this
text.
Sorry I haven't said too much nice stuff about the book, it wasn't bad,
but I think negative criticism might be more valuable than positive.
Very comprehensive, but too big! Readable yes, but oftentimes
overly wordy, and many irrelevant cliches/comments in the text and
footnotes. For me, it was best as a reference because I could look
up anything in the index and get an explanation -- that was very nice.
SUNY Buffalo Student Impressions
Excellent text book.
I like the way the book is organized.
Although book is probably the best one, it could still be better
I like the text.
The text book is really good.
I find the text book a bit opaque vis `a vis the presentation of
algorithms.
Textbook and other readings are excellent.
The book is incredible!
I have been enjoying the text book a lot! The book is quite good and
very interesting.
Columbia Student Impressions (Spring 95)
Excellent!
very readable
I liked it alot.
I liked it. Very well organized and well presented. Feels like its been
revised many times
the book pose the idea pretty easy to understand
Not bad. Pretty readable and interesting for a c.s. book. Of
course this is one of the more interesting topics too.
OK but not great.
Pretty decent book.
It presented useful information about concrete aspects of AI while
providing very interesting historical background. Unlike most textbooks
I've seen in CS, this one seemed to aim at actually relaying
information. Others usually look like texts put out by folks who like to
say afterwards, "Have you seen my new textbook?"
This book was a refreshing change.
I liked the book very much it was very easy to read, clearly written
with informative pictures and comments. It left me with a very good
impression.
Well written; useful as reference book
On my opinion the book tries to encompass too many topics.
The book had a good deal of material to present, although I did not like
the way in which ideas were presented.
Excellent.
Well-written and the student has opportunities to learn on his/her own a
lot of other cool stuff as the book is really complete.
The histories at the end of each chapter were really interesting.
Pretty helpful and well organized
I liked the book a lot. I thought it was very easy to read, easy to
understand and I thought I learned a lot. It may be though, that my
opinion is biased because of class. I would read the chapters after
class - hence I already had a pretty good idea as to what the chapters
were all about.
I liked the examples it gave...my favorite being the one on page 32,
about the "Idiot crossing the street and gets killed by a falling
airplane door". (His style of writing made me remember things more -
as well as understand his points.) All in all, I think he had great
examples and incorporated them nicely with the "real" stuff that we
had to learn.
Overall, I am very satisfied with the textbook and would
recommend it for future semesters.
I found the book very good. It is unfortunate that we never made it
to the interesting chapter of learning algorithms and machine
learning. However it is so well written that I cannot stop reading in it.
It was a great course, thanks for conveying so many new ideas.
I really liked it. Its very informative, but still quite readable.
Definitely one of the best textbooks I've had, period. I especially like
the history chapter, although it's not really applicable to the course.
Worth the money. :)
The book was one of the first really well-written books that I read as
part of the CS curriculm. I have found people in the CS
community--present company excluded of course--to be lacking in good
communication skills, and the book was refreshingly different in that
respect.
All the material we covered in class was there, all understandable
before or after class. I used it for another class's open book midterm.
Good senses of humor. Ideas are illustrated well, tangents are in
moderation and always either relevant or interesting.
The way the chapters relate to each other and the theme of intelligent
agents develops from one chapter to another makes the book flow well and
makes the entire field seem more organized than I had ever imagined.
Columbia Student Impressions (Fall 95)
10's across the board for me, Professor!
I think it's the best Comp Sci book I've bought. Thanks for making me
buy a book that's actually worth its ridiculous price!
Comprehensive, easy to understand.
some crucial ideas and algorithms not
detailed enough
The book is very enjoyable to read. I often found myself laughing or
smiling at their examples and comments. The one chapter on searching is
very dry (I think it was chapter 5), but I don't think there is a lot you
can do to make that type of material very exciting. Overall I liked the
book and I do not plan on selling it back.
The book is very good in general especially because it does have a
lot of examples, and it explains a lot of algorithms informally, making
them very understandable.
1. Effectiveness in teaching AI (0=ineffective, 10=very effective):
4 - good material, but basically overkill for me - I looked mostly
for code fragments to explain lisp, and found that this was
feasible only some of the time. As well, his examples were
less accessible than I might have wished.
2. Usefulness as a reference book (0=useless, 10=very useful):
3 - same reasons above. Utility might increase in a later class
with more advanced topics.
3. Overall readability (0=unreadable, 10=very readable):
4 - what I had time to read was readable, but not adequately
directed toward what I needed in order to accomplish the
work.
4. Overall level of interest (0=boring, 10=fascinating):
5 - pretty interesting, except that I did not have time to spend
browsing or working on understanding some of the principles.
5. Overall level of technical difficulty (0=easy, 10=impenetrable):
5 - with some lisp experience, easier to comprehend.
6. Your overall impression of the book:
5 - I feel that it may be useful in the future, but I somehow wish
that there had been a text which gave me a little more of what
you will no doubt refer to as "spoon-feeding." With a little
more low-level hand-holding I feel that the text might have
served a more important role for me in the course.
Good book! But may not suitable as a text book, but it contains too
much materials to be coverred in class.
Any text that quotes the "Witch Burning" syllogism
from the "Holy Grail" is tops with me!
For the most part, a pretty good book. Its pretty fun to read.
I think the book is very good.
It explains things well. It seems to be more functional then
technical. More technical examples would be helpful.
of all the text books i have been subjected to, this is one
of the better ones.
Harvard Professor Comments
For Harvard, the book was rated overall and got a 4 on a 5 point
scale, which we translate as 8 on our 10 point scale. Prof. Barbara
Grosz writes: Your book got a 4; previous texts have been in the
2's or very low 3's. The distinction is significant. As you know I am
a fan and think the book rates highly on all of [the questions] 1-4,
and is at the right level of technical difficulty. ...
It's a pleasure teaching from your book.
Oregon State Student Impressions
Very nice -- I was impressed with how much good information you were able to
fit into a single book.
It reinforced my enthusiasm for the field by explaining it in a clear and
coherent way.
Very good (excellent)
Pittsburgh Student Impressions
It was very good
good
!
good ... a little wordy ... good examples
I liked the book but the biggest problem was with the Lisp programs.
I had never programmed in Lisp before but was able to limp through the programs.
I had particular problems with questions 13.3 (do we encode spare tire
also seemed to be two distinct tire problems) and 8.15-8.16 (difficult should
we get the response exactly or should we just ask yes or no questions).
things I like: the index is very good ... table of contents is good ... the bold print
for keywords nad in the index for some pages helps find things.
very much miss: answers to some (alternate) questions is handy when
studying on our own
book itslef: the spine started to split
10
9
9
7
Online Questionnaire Impressions
A reasonably good text book. Although in some section like Probability
and baisian networks some solved examples would have helped to get the
ideas across mauch easier. Overall i enjoyed using it.
Execllent text,I used it for U of Michigan'n EECS492 this semester
The book was a fairly good, understandable introduction to AI. As a general reference, I didn't feel like it was specific enough, but I suppose that my opinion on that could easily change with time as I do not know what I will expect from a reference piece in a couple of years. It was usually an interesting text, the main difficulty that I had with it was in following some of the sections that were a little more math-intensive.
I thought that the book was a great benefit to the class. I think
the aspect of the book that I really enjoyed were the historical
tidbits. And that Python quote was something else. It really
helps break up the monotony of late night reading. With respect
to the question about this being used as a sort of "reference"
book, it has all of the information needed, but in my opinion
the key features of a reference material are both the information
and the structure supporting it. If you went for more of a
referential structured style, I think you would lose some of
the readability, which is what I thought made the book something
special. Keep up the good work-
Most of the areas were covered and explained well, but some aspects
were not explained to a satisfactory level (e.g. nural networks).
since this is an introductory course in AI. I think it would be
a good idea to explain some aspects from a more basic level, other
than that, I found the book to be very helpful in my course work.
Appears to be a very comprehensive introductory-level text.
One of very few course textbooks that I am likely to
read after completing the class.
Seems to do a good job of avoiding excessive technical detail
when possible.
It is very wordy. I shouldn't comment on difficulty, because the
level of difficulty I experienced was surely influenced by what
questions the professor chose from the text. It is a fine book
except for being so long-winded. It is certainly NOT a reference
text.
A well collected overview of the AI field
In places where results from calculations are presented, the steps
of the calculation should be shown as much as possible -- most
specifically in the chapters on belief network inference
calculations.
I liked the book very much. If I had to make a suggestion to
improve it, however, I would suggest that you work out more
examples more thoroughly. Some code along with the book would
also be nice.