 |
|
Title: Fluid Concepts And Creative Analogies: Computer Models Of The Fundamental Mechanisms Of Thought
ISBN: 0465024750
Author:
Douglas R. Hofstadter
Publicate Date: 1996-03-21
Publish: 1996-03-21
List Price: $24.00
Average Customer Rating: 4.0
Format: Paperback
 |
|
|
|
|
Amazon Lowest New Price: $16.00
Amazon Lowest Used Price: $5.99
Amazon Merchant Price: $21.60
|
|
|
|
|
|
|
|
|
|
|
| Customer Review: |
|
1: 15 years on, still relavant. We have a long way to go....
Hofstadter provides effectively a series of articles published elsewhere, edited in his engaging, verbose style.
Basically the question of the book how would a computer solve the following:
"X:x as Y:?"
You can get much more complex, but basically his group spents the 80's and early 90's researching this questions and trying to figure out, "know when to break the rules" applied.
His overall appraisal of AI is that even within confined realms, it still produces inconsistent results, and there is a long way to go.
Processing power is ~1000x greater than when he wrote this book, but as he observed with Deep Blue, "Brute force methods tell us nothing about Human thought".
I realized this was a small sampling of the issues facing the whole approach. Enjoy.
|
2: Another piece of the puzzle
When I first starting reading "Fluid Concepts" I found myself puzzled; what, exactly, was Hofstader up to? He and his team of grad studenst seemed to be spending a tremendous amount of time on something that at first struck me as very trivial- solving puzzles of the "what number comes next" variety. I didn't see the connection to cognition. I put the book down for a while.
When I returned to it, after having done some refresher reading in cognitive psychology, Hofstaders' intent was much clearer. To understand his program, you have to start by discarding GOFAI ideas about the stored representation being primary, and look at the problem as a psychologist would: Before you can even ask how representations are stored, you have to ask how they got there in the first place, and that's what Hofsatder is looking at here.
Perception consists in large part of taking a mass of sensory data, and looking for patterns- in it. That's a critical part of cognition. It's both how we extract words from marks on paper or sounds uttered by another, and why we see a face when we look at a full moon, or a stain on a curtain, or a piece of burned toast. Hofstader and his team are looking for those fundamental processes that allow to both match raw perceptual data to representation, and to generate those representations in the first place.
Since the publication of this book he's moved on to another research program, and having been away from the field for over a decade, I'm not sure how influential it has been. But as far as I can tell, no one else has done as in-depth an analysis of this sort of primitive pattern matching, and for that reason alone, I think it's a program that every cognitive scientist should familiarize themselves with to some degree.
|
3: A serious read for AI wonks
I read this book when it first came out. At the time I had a deep interest in all things AI. The book presents Dr. Hofstadter's experiences (along with those of his graduate students) of implementing creativity modeling systems (and others) at the Fluid Analogies Research Group (FARG). The book is not an easy read. The reader will need to be diligent and not get deterred. The book also is a bit dry in areas, but those who are truly interested in the subject matter will not mind, much.
|
4: Too distant from my usual routes ...
Many books by D. Hofstadter are at the top standings of my personal parade, but in reading this book I found myself very likely too distant from my usual interests and preferred styles. The initial part is very interesting, but when the author carries on detailed descriptions about programs' features in conversational shape, I have been quickly bored, and I have given up attentive reading turning to an eagle eye approach. I would have been by far more comfortable with a more formal explanation, because, once I make the effort to follow the thourough description of what and how a program does, it is more convenient to study its algorithms.
So, the book is surely very pleasing for people professionally involved in semantics, but I am not confident in its general interest.
|
5: Wonderful but quite dry in parts
This book is, as others have commented, different from DH's other more entertaining books. It is a serious attempt to discuss the real issues and difficulties with AI research. There is a lot of quite dry material and in places it is repetitive. It provides terrific insight into the problem of imitating human thinking at a deep level, and I found it very rewarding. It was also very interesting to follow the threads of how he went about doing research, and what he thought of other AI research. His views of various flavours of AI research were very instructive and inightful I thought. In summary a good book, but this is not (high quality) brain candy like Godel Escher Bach etc.
|
|
|
|
|
