Firstly, I must admit I’m a bit of a Kurzweil Fan. Not in a groupie kind of way, but more in a way that I’m an optimist who believes in technology, and so is he. So as soon as I found out that he’d written a new book (which was about a day after it was released) I ordered it. Then on Monday I had and started reading… I’m not a speed reader and have no intent to even try on this book, so it will probably take me until the weekend or beyond to finish it (I also have a life that doesn’t involve computers & books ;). I therefore thought I’d to an Agile / Iterative / Incremental / Progressive review. I won’t publish every day, but every few until I’m done (and I may slow down near the end as I suspect it’s one of those books that builds up as it goes).
My plan is to go by the chapters I’ve read, then summarise / comment and probably go back over the whole thing and refactor in to a “Gold Release”.
Before we get in to my review, for a counterpoint, you may want to check out The New Yorker’s “Trendy, Dubious, Sarcastic” Review (just my opinion ;) entitled Ray Kurzweil’s Dubious New Theory of Mind. It seems to be not just me that has this opinion though as one of the Anonymous comments best sums up reasoned (to me) opinion:
This is a very poor excuse for a review. It misses so many points that it appears to be constructed for the purpose of allowing Gary Marcus to pose as a smug intellectual superior to Ray Kurzweil, and the ad hominum crap in the comment section only confirm my dismay at how thoroughly personal psychology trumps reasonable thought. Demeaning Kurzweil’s understanding may get you off, but it’s clear to this reader that this book is filled with interesting information and insights, and is a valuable contribution to the intellectual life of anyone interested in these matters.
I mean, who knows who Gary Marcus is and what has he done (other than write a seemingly vengeful review) in the bigger picture vs Ray Kurzweil?
Anyway, on to the good stuff – the review!
As I mentioned, I’m quite familiar with Kurzweil and the history of science, so in this light I was personally a bit “disappointed” by Chapter 1 “Thought experiments on the world” and Chapter 2 “Thought experiments on thinking” as I already knew most of this and therefore skimmed quite a bit. I can understand why it’s there though as this is an “enthusiastic layperson” book and he has to give some background. For me, Chapter 3 “A model of the neorcortex: The pattern recognition theory of the mind” is where the book really starts – it’s a great introduction to Kurzweil’s theories and obviously a basis for the whole book that will be expanded on. It talks about his theory that the brain operates as a hierarchy of pattern recognisers with some illustrations. Part of this theory is that the brain works in one-dimensional lists – what a mind blower, as this is the basis for Lisp which was invented by John McCarthy. There is much reference to his earlier work, such as the “Book reading machine for the blind” and the “Kurzweil Synthesiser” (which I’d buy if I was loaded) that shows this is the track he’s been on all along and is finally realising it…
Now we start to get down and dirty and gooey as Kurzweil starts off in Chapter 4 covering “The Biological Neocortex”. This was really interesting to me as my knowledge of the brain and it’s various areas is quite basic. He also refers to the recent NIH study of the brain and had the same thought as me – it looks like a crossbar switch! Are we inherently creating brains in our technology? The next Chapter 5 covers “The Old Brain” where Kurzweil contends the Neocortex has taken over much or at least severely enhanced many Old Brain functions. Hormones are given pretty short shrift though – maybe Richard Berglund, author of Fabric of the Mind may disagree. Only time and experimentation will tell. This is all pretty much wrapped up in Chapter 6 “Transcendent Abilities” which covers Love and Aptitude that touches on nature vs nurture but kind of neatly sidesteps the whole issue.
Having set the biological background in the previous chapters, Kurzweil now moves on to the question of “how do we build one of these brain thingies”, starting with Chapter 7 “The Biologically Inspired Digital Neurocortex” which starts off talking about brain simulations, gives us some more background on his original research and outlines Vector Quantisation that is a technique that provides a compressed way to capture learning. He also outlines Hidden Markov Models which seem to have a similarity to the brains functions. There’s also mention of LISP, which also works in 1-D lists – coincidence or reflection of our structure? After all that theory, there’s a nice general discussion about out current efforts in machine intelligence, such as Watson which leads nicely in to Chapter 8 “The Mind as Computer” that gives a compressed history of the original development of computers. My only criticism is that like most Americans he overlooks Turing in preference to von Neumann with regards to implementing a fundamental computer architecture. There is one thing I didn’t know here, which is that von Neumann wrote near the end of his life about neural processing and even tried to estimate the amount of computing needed – these estimates still hold today!