From rec.arts.sf.reviews Thu Nov 30 16:56:37 1995 Newsgroups: rec.arts.sf.reviews Path: news.ifm.liu.se!fizban.solace.mh.se!news.ecn.uoknor.edu!paladin.american.edu!howland.reston.ans.net!newsfeed.internetmci.com!news.kei.com!uhog.mit.edu!news!nobody From: Jonathan Thornburg Subject: Review of Harrison & Minsky: "The Turing Option" Message-ID: Followup-To: rec.arts.sf.written Keywords: author= Jonathan Thornburg Sender: news@media.mit.edu (USENET News System) Organization: Date: Tue, 28 Nov 1995 03:25:28 GMT Approved: wex@media.mit.edu (Alan Wexelblat) Lines: 107 [some minor spoilers --AW] This is an interesting, though flawed, novel about the attainment of human- level artificial intelligence (AI), set in the 2020s. Minsky is well-known as a major AI researcher and former director of the famed MIT AI Labs. Accordingly, the book is free of the glaring technical howlers found in many fictional discussions of AI. The writing, plot, and characterization are fairly good, and the book met my personal litmus test of keeping me up past midnight the first night I brought it home from the library. But the authors reaffirm common stereotypes about the personalities of gifted children and adults, the sociology of scientific research, and the dehumanizing implications of AI. The plot revolves around a brilliant AI researcher, Brian Delaney. Just as he's making a major breakthrough in his AI research, unknown persons penetrate the high-security corporate lab where he works, and shoot him in the head (this is the _United_States_: there's a fair bit of blood and gore!), leaving him for dead. They then steal the entire contents of the lab, including all the records of Brian's work. (Just for good measure, they also remove a goodly number of corpses of assorted guard dogs, guards, and other corporate personnel caught in the attack.) However, Brian survives the attack, and his employers recruit a top brain surgeon to try and repair Brian's badly injured brain, in the hopes of his being able to either identify his attackers, or better yet recreate his AI breakthrough. The surgeon successfully implants a number of computers and "nervous system switches" in Brian's brain, and Brian regains conciousness in relatively good physical condition, but with much of his memory and personality -- his "mind" -- still shattered. Working together, the surgeon and Brian successfully program Brian's brain-implant computer to reconnect the severed nerves in his brain. The "bionic" Brian then sets about recreating his AI research, and eventually succeeds in constructing a true "machine intelligence", far more capable than the prototype stolen in the attack. Unfortunately, Brian's original attackers are still at large, and they see Brian and his work as a threat to their AI monopoly. Brian survives several further attacks on his life, but is forced to live a secluded life in an isolated high-security laboratory. He eventually rebels against his imprisonment, and escapes to try his own hand at outwitting his enemies. I won't reveal the remainder of the plot, but the ending is both interesting, poignant, and disturbing. The basic concept of the book as a good one: What are the implications of successful AI -- of a computer system which is fully as intelligent and self-aware as a human? But I think the authors fall down in several areas: Brian is presented as an exceptionally gifted child, who (for example) discovers the concept of "prime number" for himself at the age of 8. But the authors' presentation of the adult Brian -- and later the "bionic Brian" -- as a lonely workaholic, incapable of meaningful adult relationships, owes more to TV stereotypes than to reality. In fact, although gifted children often lead rather unhappy childhoods, they generally emerge as adults who are psychologically _healthier_ than average. The authors should have known this. I also thought the authors presented several of the characters as pretty naive: When breaking out of a high-security military compound, would Brian and his accomplice really admit into their conspiracy, the military's nominee as Brian's research assistant? Is it really plausible that such a person, presented as an AI expert with an extensive computer database background, wouldn't know that cellular phones _must_ continuously broadcast their position to the telephone network, in order for incoming calls to be rounted to the proper cell? And should both of them, knowing they were fleeing the entire resources of the US government (not to mention some other very well-bankrolled enemies), go about using credit cards, passports, and the like all _in_their_own_names_? The authors present Brian as working alone, with noone else in the world being able to follow his work. This is in highly improbable: new scientific discoveries almost never come as bolts out of the blue, but rather out of experience with the limits of existing scientific knowledge. Accordingly, there are almost always a number of researchers around the world working along similar lines. Moreover, most scientists work in groups of anywhere from 2 or 3 to as many as 1000 researchers. Again, the authors should have -- and Minsky surely does -- know this. Finally, let's consider the central philosophical subject of the book, the nature of AI and its impact on humans. Here Minsky's AI background shows: the writing is authoritative and the authors carefully avoid some of the common misconceptions in the field, and even debunk a few others. But although they use Alan Turing's name several times, and even have a short preface titled "The Turing Test", they never actually explain the Turing test's central roles in AI. The ending presents Brian's newly-created AI is growing steadily more human, while the "bionic Brian" himself becomes cold and impersonal. But the authors neither justify this divergence nor expand on any of the other sociological impacts of AI. Overall, then, this book is a "good read", but I expected much better from someone with Minsky's background. I'd give the book a 6/10: don't buy it now, but perhaps check it out in your local library, and keep an eye out for a 2nd-hand copy. %A Author Harry Harrison %A Author Marvin Minsky %T Title The Turing Option %I Publisher Warner Books %C Place Published New York %D Date Published 1992 %G Order Number ISBN 0-446-51565-5 (hardcover) %P Pages 422 pp - Jonathan Thornburg U of British Columbia / Physics Dept / "There is nothing to it: You only have to hit the right note at the right time, and the instrument plays itself." - J. S. Bach