Customer Reviews

After Thought: The Computer Challenge To Human Intelligence

5 star:		(4)
4 star:		(5)
3 star:		(0)
2 star:		(0)
1 star:		(2)

 

 

I'm an artificial intelligence person, and I bought this book to keep up-to-date with what popularizers are saying about my field. Now I wish I hadn't--it's too depressing. James Bailey got caught up in the promise of the wave of trendy data-driven learning algorithms like neural networks and genetic algorithms, and the purpose of the book is to gush about them. There's no attempt to place these methods in the broader context of artificial intelligence as a whole, and no recognition of their weaknesses. The book randomly names these methods the "new intermaths" and analyzes them historically as a radical breakthrough from the traditional sequential, numeric, data-poor human reasoning into the brave new world of parallel, non-numeric, data-rich silicon reckoning. He's partly right about that, but I think it's a coincidence; the book's argument is so muddled that it doesn't reach any convincing conclusion.

Bailey's ruminations bear rereading now that five years have passed since their first appearance. We have survived the advent of the new Millennium, and we have found once again that, after the divine afflatus, human intellect reigns supreme in the cosmos.

This is not to say that Bailey's postulations have no merit. The advances of the computer age, particularly in the still-infant and arcane discipline of artificial intelligence, continue to fascinate us and to challenge us. Yet even with the burgeoning networks and the increasingly powerful integration of humanity and its machines, the surprises offered by the observations of the young and young-in-spirit still outdo those of scientific teams. I suspect that pattern, as old as civilization, will prevail.

The chapter in this book that most demands reflection-and rereading-is Chapter 17, where Henry David Thoreau is pitted against the myriad forces of the information age. As early as 1978 I was criticized by my literary colleagues for teaching the metaphysician of Walden Pond. At the time, I was working with Dr. Louis Uhle of USC on patterns of word usage in Renaissance English (primarily dramatic) texts as a scientific measure of authorship attribution. Yet Thoreau offered my students access not to data, but to genius, and that, not data, intrigued them.

Another colleague from the life sciences took his students to the college lawns to look as if for the first time at the patterns of dandelion blossoms, evoking the kinds of questions that not much later spawned Chaos Theory. My maths colleagues, intrigued with what I was doing with language, proceeded to AI, and my psychology colleagues drifted off to work some of the same ideas in formulating new network designs at Bell Labs.

Bailey's own accounts of elementary school children discerning the grounds for identifying a species that had remained unperceived by the "experts" reminds me of those halcyon days-before I also drifted into "line and circle" problem solving of the probabilistic kind. We had no idea we were participating in a revolution of thought, and perhaps that was just as well.

A metaphor Bailey uses in After Thought, particularly in Chapter 17, is that of understanding the behavior of rivers, and I think he was trying to suggest something about the elusive construct of Nature, which, if we should only drop our preconceptions and listen, would always surprise us. The Mississippi, like the Nile (or the Yangtze, for that matter), remains a keen scientific concern, but offers no easy understanding.

Yes, we have the potential to engage in collaborative enterprises involving computers and networks around the globe-the greatest parallel processing enterprise, in size and scale of any age. The question is, how much closer to essential truths this endeavor brings us?

Bailey would have us understand that we are about to transcend the time of maths as we know them and that we shall reach a new plateau of pattern recognition that renders the schema of ancient Babylon, the thinking of Kepler and even the cogitation of the Sante Fe Institute fellows obsolete. Perhaps, but, then again, perhaps not. The leaps of genius have outrun the numbers throughout history, however the scientific elite have formulated them. Our experts propose; Nature disposes otherwise. The human intellect itself has been found far more powerful than the sum of all the crunching power of all the machines in being or under consideration-working individually or working together.

Two cautions to Bailey's line of reasoning-for its determinism seems to partake of the confining features of Newton's laws, which he claims to eschew. First, no true scientist ever deceives himself or herself that absolutes are inviolate. Second, the very open inquiry at the root of the scientific method-even among the ancient Greeks-has never really surrendered to the rigors of numbers.

We fall between the pattern that has broken and the patterns that we sense as possible in Chaos every time. Taken that way, we can all still engage in our common task of understanding taking the fruits of our data looms as yet another set of features against which we try to move from apprehension to comprehension, the latter of which we approach as that dark and all-transforming glass of eternity.

This is a wonderful intellectual history of modern logics. However, Mr. Bailey should probably have known more about brains and less about computers before he wrote this book. That human brains can think sequentially and that this has been the mode of scientistic thinking since Descartes, does not mean that brains only think sequentially as he assumes. Furthermore, that computers can be built in parallel does not mean that "wires" have any better capacity to process intermaths than brains. His conclusion, therefore, that we need parallel computers to complement our serial brains is silly. Alas, he is as captivated as all of the busy complexity theorists at Santa Fe by the notion of the "object" whose interior processes can remain "intractable" so long as we recognize their behaviors. A better conclusion might have been that since we can see computers producing order without bendfit of formulaic algorithms, then perhaps we should investigate if this isn't really how the massive parallel processors between our ears actually work. Or, he could spend some time just observing and reporting on the creative processes of ... well ... artists

James Bailey boldly articulates the new communications paradigm for the bio-information age. His fascinating account of the evolution of human thought and its present transition to relying on the "alien logic" of machine parallelism reaches beyond the history of human socio-evolution to the very fabric of our existence as intelligent beings. This book is an A++++. I reccomend it to all sentient beings.....

After Thought delivers a convincing argument that humans think in patterns that optimize biological life -- slow "switches" and massively parallel processes in "meat processor." New silicon life, on the other hand, can choose other routes. Rather than using traditional mathematics (equations), computers may discover truths by massive databases or other unknown techniques. The author gives the example of how our intellectual paradigm changed from geometry in classical times to equations after the renaissance. We are ready for a new paradigm and nonbiological or semi- biological machines may supply the solution. For some, however, this is an unnerving development. Solutions and knowledge may be developed which we (humans) cannot trace to "first principles" or even follow. Good read for those who like to think. If you read the book "The Man who fell to the earth" you'll remember one key point made by the alien -- "On earth you only had one intelligent species." Maybe that is about to change. Bill Yarberry Houston, Texas

This is a delightful book! Bailey brings a historian's eye to an intellectual history of mathematics, and how our tools for calculation shape the things we calculate. Every page is full of details --- Galileo's massively-parallel thought-experiment with beads on a wire; how the shift from hand-written manuscript to movable-type printing press sped the replacement of geometry with algebra; Kepler's lifetime of calculation; the intriguing suggestion of how different was Babylon's astronomy, which was not based on geometry. Bailey sees the history of mathematics filtered through three phases: the mathematics of place (geometry), the mathematics of rate (algebra to calculus), and now, with the emergence of computer tools, the mathematics of change and relationship (chaos theory, neural nets, simulation).

This book had a profound effect on how I view a number of subject including AI, Evolution and Medicine. I see a review by an AI guy who was disappointed and I have to agree with that perspective. If you are looking for a how to book don't buy this one. If you are looking for an intro to AI book I personally like Turtle Termites and Traffic jams but I would not call it a how to book either. But to me After Thought is profound. I feel that anyone who is interested in learning ideas behind emergent intelligence or is interested in fields that may involve emergent intelligence should read this book.

Bailey does an excellent job of combining computer theory with history, mathematics, economics and philosophy. I can only guess that the people who gave it a single star must have had difficulty with Bailey's style which, at times, requires a measure of diligence to locate the point.

This book is now over 12 years old, yet it is still worth reading.

The author tries to show that there are 3 eras of how we tried to codify nature - the period of trying to define "place", e.g. the orbits of planets, the era of "pace", e.g. the use of equations to show how things changed with time, and finally, the modern era of "pattern". Each period shaped how we think about the world and Bailey clearly thought that pattern was the next big thing. In some respects he was right, and in some wrong.

He was right in suggesting that new techniques, such as those that are now called "machine learning" would have a much bigger impact on data analysis. They are certainly more used, but never as prominent as he and others have suggested they would be. He was right that we would be facing a deluge of data. However he was wrong to think that computers would handle it all. As it turned out, humans still do most of the work, and the internet has made that a lot easier. Some examples do exist of where the machine does the analysis work, most notable the SETI@home project and others like it. But in other cases, such as the ubiquitous "Capcha", it is humans who are using their pattern recognition capabilities to do "mechanical turk" work.

As with other books of this kind, there is a quaint, other wordly quality to the ideas as events have moved on and new patterns of computing and organization have developed. The earlier chapters are the best, as they are embedded in history. The ideas in the pattern era chapters are now dating rapidly. although they do exhibit a certain prescience in the description of where scientific analysis methods might go.

Who would have thought that at the time of writing, Google was analyzing hyperlinks from millions of web pages to create page rank data for web search and would become one of the greatest new era companies, parsing and analyzing billions of web pages today, proving a wealth of data on web structure and usage.

Overall an interesting read that will crystallize some ideas about how we have progressed in our understanding of the world and where we might go. This is more a philosophical work suitable for a generalist reader, rather than someone versed in teh fields of computational analysis looking for clues on what is new and worth pursuing.

Thought has been humanity's greatest claim to fame since Descartes linked being and thinking in his famous first principle. Before that we knew we were cool because God parked us in the middle of the universe. At least we thought that's where we were, until the Italian joker Gallileo Gallilei picked up on Copernicus and Kepler and knocked out our stilts. James Bailey believes computers are now poised to swipe our mental crutches, to move thought to a higher level, an alien place where we cannot follow. He argues convincingly that we are now entering a period of change unlike anything since heliocentrism cleaned our clocks four hundred years ago. Bailey divides our intellectual history into three phases (so far). At first we were concerned with Place, then Pace, and now Pattern. The Greek geometers aimed to calculate where they were, track the moon and planets, and predict eclipses. They soon discovered that the world was round. When curiosity hooked up with technology which made long seagoing adventures possible, sailors wanted a reasonable shot at getting home. Navigation demanded better math, and soon algebra led to Newton and the calculus. Newton's mental world is the one we inhabit, but the looming questions of ecology, biology and behavior are beyond the ken of Newtonian physics. The first fifty years of electronic computing have consisted of letting circuits do what we do, albeit faster. The search for "artificial intelligence" has really been an attempt to create machines that follow Cartesian logic and work through equations sequentially. That was then, this is now. Sequential thought is just fine if you have limited data and need to go from A to B to C. Parallel thought is much better if you have a ton of data and want to compare every permutation. Notice that we are awash in information these days. (If you think you are keeping up you are obviously confused.) Computers are now being designed that use thousands of parallel processors. It's sort of a hypercase of two heads being better than one. Programs are no longer static instruction sets, instead they are allowed to evolve. Intermaths are developing that can expand genetically and learn to solve problems in ways that we cannot hope to fathom. All we can do is trust the programs that consistently offer right answers. Don't ask why. Don't ask how. The Internet is poised to become the biggest parallel processor of all with millions of computers able to access jillions of bytes of data. Hang onto your mice, folks. AFTER THOUGHT is a fascinating look at thought, computers (human and electronic), clocks, printing, science, time and your sense of self. Parts of the book are even written in parallel. It is well worth a bit of your limited time to find this intellectual raft in the vast sea of print and sail it on home.