Kurzweils argument for the success of AI》.pdfVIP

Artificial Intelligence 170 (2006) 1227–1233 /locate/artint Book review Ray Kurzweil, The Singularity Is Near: When Humans Transcend Biology, Viking Adult, ISBN 0670033847, September 2005, 672 pages. Kurzweil’s argument for the success of AI Drew McDermott Yale University, Department of Computer Science, PO Box 208285, New Haven, CT, USA Available online 7 November 2006 “Positive feedback” is one of those phrases that sounds like it ought to mean something good, when it actually means something that is almost always bad: a runaway process whose output forces its inputs further in the direction they are already headed. A chain reaction in an atomic bomb is a classic example, where the production of neutrons is proportional to the number of atoms split, which is in turn proportional to the number of neutrons. Of course, a bomb is one of the few cases where the goal is to produce a runaway process. Even in this case, positive feedback has the property that it quickly destroys the conditions that made the positive feedback possible. This is a hallmark of the breed. Take any other example, such as the multiplication of rabbits, whose output—rabbits—is proportional to the number of rabbits. This simple model assumes that food is abundant; the exponentially increasing rabbit population soon violates that assumption. In Ray Kurzweil’s book, The Singularity is Near, it is argued that the evolution of humanity and technology is driven by a positive feedback, and that the resulting destruction of many of the truths we hold dear will result in a post-human world of wonders. It’s hard to argue with the claim that a positive fee