Let me try to help you. Many people who praise GEB in the highest terms and recommend that everyone read it never finished it. Many read all the dialogues, but only some of the chapters. I have absolutely no data to support turning either of the previous "many" to "most", but wouldn't be surprised by either possibility.

GEB's most important strength, by far, is in giving you a diverse set of metaphors, thought-patterns, paradoxes and ways to resolve them, unexpected connections between heretofore different domains. It enlarges your mental vocabulary - quite forcefully and wonderfully if you haven't encountered these ideas and metaphors before. It's like a very, very entertaining and funny dictionary of ideas.

The exposition of various topics in theory of computation, AI, etc. that it also contains is not as important by comparison, and isn't the best introduction to these topics (it's still good and may well be very enjoyable, depending on your background and interest).

So there's no reason to fear reading GEB. You'll chuckle with recognition at the jokes, metaphors, notions that you've already learned elsewhere, and will be delighted at those you've never seen before. Read all the dialogues; if some of the chapters bore you, resist guilt tripping and skip a few - you'll come back to them later if you need them.

I thought GEB was a very good book, but much of it consisted of things I had learned about up from several other sources (formal logic, Godel's incompleteness theorem, and some others) and I also found reading it to be slow and effortful - not the kind of book that keeps me compulsively turning pages, hungry for more. I also don't know how well it conveys its concepts to someone who never heard them before, because I already understood many of the concepts that GEB tries to painstakingly explain to an audience of people who hadn't already taken three university courses on formal logic. (And I did read every last word of it!)

GEB has been described as "an entire humanistic education in one volume" and that's not far from the truth; I could easily imagine using it as the textbook for several consecutive semesters worth of university classes. It starts with some basic concepts that the proverbial smart 12 year old could understand, and builds on them, step by step, until the book is covering concepts and applications on the level of graduate school. The book is obscenely ambitious in its scope - imagine a single book that assumes the reader has never taken a course in algebra and wants to teach that reader enough that, by the time he or she reaches the end, can understand and discuss university-level calculus, and you'll picture something much like GEB.

So, yeah, GEB really is amazing, but it's also a headache-inducing monstrosity that will try to cram your head with concepts until it explodes.

I agree with some of the sentiments below. Also, I became a philosopher before reading GEB, and didn't really find anything particularly enlightening in it. I still recommend "The Future and its Enemies" much more highly, and it's more of a fun read (even though it's a bit dated now)