HAL's Legacy: 2001's Computer as Dream and Reality. - book reviews

Edited by David G. Stork. Foreword by Arthur C. Clarke. Cambridge Mass.: MIT Press 1997. 384 pp. Hardcover, $25.

I remember seeing the film 2001 for the first time in 1969, at a Boston movie drive-in, in the back of my parent's station wagon. As I recall, my brother and parents fell asleep, and I stayed up and watched Kubrick's and Clarke's vision of an interstellar future. Now, as a very failed scientist and a somewhat successful artist, I read with some sadness Stork's dedication at the beginning of this interesting and visually beautiful book, "To Stanley Kubrick, Ars sine scientia nihil est." "Art without knowledge is nothing." As Stork and other authors in this book point out, we arrogantly predicted HAL in a movie, but have yet to actually invent him. In some ways, Artificial Intelligence such as the chess-playing Deep Blue and the computer psychologist ELIZA have come close, but we have yet to invent a machine that can actually learn and feel emotion with the depth of a human being.

David Stork is a cognitive psychologist. With his colleagues, he takes apart the 1960s fascination with computerlike characters like Star Trek's Mr. Spock and HAL, and says in essence that no machine is as brilliant as a human memory or emotion. But we have come close, if only as a result of our striving for brilliance and desire to excel.

The book is peppered with pictures from 2001, charts and graphs comparing the voices and movements of HAL and humans, showing us how far we have come and how far we still have to go. Daniel C. Dennett's last chapter takes us back to the time we invented the computer and discovered atomic physics, and asks "When HAL Kills, Who's to Blame? Computer Ethics." For those of you unfamiliar with the movie, HAL disconnects astronaut Frank Poole from his life-giving tether and pushes him out to space and death in a cold void.

Raymond Kurzweil points out that we have been able to create computers that talk with inflection, but not with emotion, that can parse a sentence perfectly and logically, but cannot master the nuances of slang and grammar that we learned in our first 12 years of schooling. In an interview with Stephen Wolfram, Stork and Wolfram discuss the popular educational math software Mathematica, and how it was distributed before HAL's fictional 1997 birthday. The book even devotes a chapter to the possibility of a computer that can lip-read as HAL did in the film.

Four of the chapters are Dr. Stork's, the remaining by science and film experts such as Arthur C. Clarke, Marvin Minsky, Murray S. Campbell, Joseph Olive, Roger Schank, Douglas Lenat, Rosalind Picard, and Raymond Kurzweil. Each chapter after the first takes an attribute of HAL, as he functioned in the movie, and asks "Could we do this now?" Storks and his colleagues' answers are a resounding, No.

The book asks intriguing questions about the distinctions between human and computer. Various attributes of what it means to be human are broken down and analyzed; Can computers mimic humanity, and if so, to what degree?

Azriel Rosenfeld asks, Can a machine see? Well, not as we can. It may see a numerical array of pixels, but not be able to make sense of it. Can a machine talk with emotion? Well, we have created the computerized voice, but still cannot invent the inflection that makes it humanly responsive. There are humanistic chapters such as Rosalind Picard's "Does HAL cry Digital Tears? Emotion and Computers." When Dave, the last astronaut aboard, figures out HAL has cut Frank Poole adrift, he disassembles his last friend in the universe. HAL, we find, has a fear of death or at least of non-existence. He pleads sadly for Dave not to disassemble him. Douglas Rain, the actor's voice behind HAL, brilliantly portrays a fear of discontinuance that is both terrified and without feeling at the same time.

Murray Campbell writes that we got the chess thing right. The ordeal of chess master Kasparov, who was finally beaten by Deep Blue, is discussed. Alan Turing also wrote a computer chess program, the book mentions, even before he invented a computer. HAL plays chess with human style and insight, even a mistake now and then. Deep Blue merely computes all possible moves of his opponent, no accounting for human error.

What is our fascination with the computer anyway? Stork and his colleagues ask. As Arthur C. Clarke reminds us in the foreward, the first computers were built for warfare in the 1940s. The Z-3 by Konrad Zuse in Nazi Germany, the Mark I by U.S. Navy Commander Howard Aiken, and the Robinson and Collossus computers by Alan Turing and his English colleagues are all examples. Stork and his associates suggest in this darkly entertaining book that perhaps the roots of the Computer Revolution were not so innocent.

After reading this book, I think back to that innocent time in 1969, when 2001 seemed very close and our HAL and mission to Jupiter were not yet here. Perhaps the year, as in George Orwell's 1984, was not really the point of their predictions. If art without knowledge is nothing, as the Latin saying goes, perhaps it is because their relativity is derived equally from the human imagination.