THE SOULS OF THE NEW MACHINESJanuary 1, 1989, Sunday, Late City Final Edition
Section 7; Page 10, Column 1; Book Review Desk
By M. MITCHELL WALDROP
M. Mitchell Waldrop is a reporter for Science magazine specializing in physics, astronomy, space, computers and cognitive science. He is the author of ''Man-Made Minds,'' a study of artificial intelligence.
MIND CHILDREN The Future of Robot and Human Intelligence.
By Hans Moravec.
Illustrated. 214 pp. Cambridge, Mass.
Harvard University Press. $18.95.
About four billion years ago, according to one theory, certain clay minerals in the stream beds and oceans of the primordial earth served as a kind of scaffolding for the creation of life. The billions of microscopic crystals in clay provided billions of surfaces where simple carbon-based molecules, such as amino acids, were able to gather and link up into more complex molecules, such as proteins and nucleic acids. In time, these larger molecules began to interact with each other, becoming more and more sophisticated until they were able to float free and reproduce themselves without the help of the clay. And when that happened the mineral scaffolding of clay was no longer needed; carbon-based life as we know it was well under way.
Now, no one can say whether or not this theory is true. But it does lead to an intriguing question: is something similar happening now? Does the ever-increasing power and autonomy of our computers and robots mean that we are headed toward a new transition, one that this time will see biological life itself superseded?
For Hans Moravec, director of the Mobile Robot Laboratory at Carnegie Mellon University in Pittsburgh, the answer is a resounding yes. If robotics and computer technology continue to develop at current rates - a prospect Mr. Moravec finds no reason to question - he predicts that robots will pass the threshold of humanlike abilities within 50 years or so, at which point they will no longer need us. While many readers will doubtless find this prospect terrifying, Mr. Moravec glories in it. The robots he envisions in ''Mind Children'' will be nothing at all like the rigid, emotionless automatons we've come to expect from bad science fiction movies. They will be our companions, our helpmates and our heirs.
''Today,'' Mr. Moravec writes, ''our machines are still simple creations, requiring the parental care and hovering attention of any newborn, hardly worthy of the word 'intelligent.' But within the next century they will mature into entities as complex as ourselves, and eventually into something transcending everything we know - in whom we can take pride when they refer to themselves as our descendants.''
No longer dependent on the slow, random process of Darwinian evolution, Mr. Moravec's robots will be free to adapt themselves to environments ranging from the ocean floors of Earth to the cloud tops of Jupiter: ''Away from earth, protein is not an ideal material,'' he writes. ''It is stable only in a narrow temperature and pressure range, [ and ] is very sensitive to radiation. . . . [ But ] before long, conventional technologies, miniaturized down to the atomic scale, and biotechnology, its molecular interactions understood in detailed mechanical terms, will have merged into a seamless array of techniques encompassing all materials, sizes, and complexities. Robots will then be made of a mix of fabulous substances, including, where appropriate, living biological materials.''
Such beings need bear little resemblance to humans, he says - or, for that matter, to C3PO and R2D2 of ''Star Wars'' fame. To give a sense of the possibilities, Mr. Moravec imagines a treelike device that he calls the robot bush. It has a rodlike body about one yard long, where many of its sensors and much of its brain power are concentrated. It also has two arms, each containing a fair amount of brain power in its own right; four reasonably smart fingers; eight fingerlets - and so on until its ever-branching limbs finally terminate in roughly one trillion microscopic appendages. ''The bush robot could reach into a complicated piece of delicate mechanical equipment - or even a living organism,'' he writes. It could ''simultaneously sense the relative position of millions of parts, some possibly as small as molecules, and rearrange them for a near-instantaneous repair. In most cases the superior touch sense would totally substitute for vision, and the extreme dexterity would eliminate the need for special tools.''
Well, all this is fun, not to mention mind-expanding. But then, so is science fiction. What makes Mr. Moravec's book different is that he wants us to take it seriously, as a realistic sketch of our not-very-distant future. And that means that a reader has a right to expect some answers to two very basic questions - answers that ''Mind Children,'' unfortunately, does not always provide.
First, is any of this (super)human robotics technology even possible? Here, Mr. Moravec is at his best. As someone who has been at the forefront of robotics research for nearly two decades, he has clearly done a lot of thinking about how to get from here to there. Some of the steps he proposes are admittedly sketchy. Nonetheless, in the early chapters of ''Mind Children'' he has given us a comprehensive and highly readable survey of the state of the art in robotics. He includes detailed discussions of sensors, grippers and mechanical legs - and why such ''simple'' things as walking and seeing are so much more difficult than ''hard'' mental tasks such as playing chess or proving mathematical theorems. The book is worth reading for these sections alone.
However, I was left feeling curiously unsatisfied on one count. Mr. Moravec's whole thesis presumes that sufficiently advanced computers and robots will possess consciousness in the same sense we do; otherwise, how could they become our heirs? And yet, except to say that consciousness will be ''emergent'' in complex systems, he does little to justify this presumption. In fact, it has been the subject of many a debate among philosophers, psychologists and artificial intelligence researchers: is thought just a form of (very complex) information processing? I happen to believe it is. But there are a lot of deep issues here, and I think ''Mind Children'' would have been stronger with a much better discussion of them.
Leaving that issue aside, however, we still have the second basic question: is Mr. Moravec's vision of the future desirable? And here, I think, he falters. Nowhere in ''Mind Children'' does he express the slightest doubt that what can be done will be done. (Indeed, he argues that nations will be forced to develop robots as fast as possible in the name of economic competitiveness.) He says very little about the impact of humanlike robots on human employment - or, more importantly, on humankind's sense of purpose and dignity. Apparently we are simply to fade away, leaving the universe to our heirs and betters. And he gives only passing mention to that ancient human institution, warfare. The fact is that military needs have always been among the major drivers of computer and robotics research, both here and abroad; Mr. Moravec's near-human machines might very well find their first uses as robot soldiers and robot fighter planes.
Perhaps it is true, as Mr. Moravec argues, that none of those doubts will make any difference in the long run. Nonetheless, without a better and more sensitive discussion of these issues, ''Mind Children'' comes perilously close to the kind of uncritical gee-whiz that gives technological optimism a bad name. I happen to think that Mr. Moravec's vision of our long-term future is an important one, worth thinking about as seriously as he does. But for that very reason, it deserves a better setting.