Robots long have posed intriguing possibilities; they have been presented as either beneficent servants or ominous enemies. Fictional images range from the bumbling but helpful robot from television’s Lost in Space to the insane HAL 3000 from 2001: A Space Odyssey. In the real world, robots have taken over unpleasant, dangerous work such as spray-painting automobiles, but they also have displaced human assembly-line workers, some of whom had no other employment readily available. Naturally, then, people have mixed feelings about robots.
Hans Moravec’s latest book will do little to clear those feelings but might crystallize them. He concludes that robots are the race of the future: They will displace humans because their information processing capacity will far exceed that of people, giving them capabilities we can only imagine—including the capability to develop even more advanced versions of themselves.
Moravec sees this prospect not as frightening but as exciting. He believes that humans will give way gladly to their robotic descendants, much as they now move aside for their biological children. Moravec in fact refers to robots as “mind children” and sees them as the human race’s intellectual descendants. We will give them some intellectual power through inheritance, and they will learn even more; as “parents,” humans will hope that their robotic children will achieve even more than they did.
Moravec is clear that he views robots as having a type of intelligence and as living beings. To get to that point, he offers a history of robotics paralleling the development of human civilization. He sees a first generation of robots as developing spatial sense (at present), to be followed by a second generation with adaptability (about 2020), a third with imagination and problem-solving ability at the level of a monkey (about 2030), and a fourth with humanlike reasoning powers (about 2040).
Chapter 1 outlines the advance of human culture and technology. With the advent of written records came rapid advances in culture, more rapid than advances in human biology. Work, as a cultural artifact, advanced beyond biological adaptation, becoming less natural and more artificial. Work requires increasing amounts of training because it is not biologically based, and workers naturally feel disconnected from their jobs.
That will change in Moravec’s imagined future. Robots will take over “unnatural” jobs and most production work, leaving humans with the options formerly held by the idle rich. The human race can return to a more natural state, with personal relationships restored and time freed for intellectual and artistic pursuits. Moravec sees these developments as “chaotic,” meaning that they are very sensitive to minor changes and thus unpredictable. “Tame” machines will serve and protect the human race, but unpredictable “wild” ones, with capabilities their programmers did not foresee, eventually will take over.
Moravec grounds his predictions in the history of robotics. The earliest robot vehicles, from the 1940’s, could do little more than navigate hallways and find defined objects such as light sources or wall sockets. The 1950’s brought a limited form of artificial intelligence and machines able to solve elementary problems; they had little ability to generalize and lacked “common sense.” Reasoning power expanded greatly, and 1997 brought the defeat of Garry Kasparov, the world’s best human chess player, by a computer called Deep Blue.
Moravec was part of pioneering work in robotic motion and vision in the 1970’s, and his experiences provide perspective on the problems faced by robots. Humans have had thousands of generations to evolve the ability to interact with their physical world; robots are far more primitive, and their evolution, guided by humans, has favored intellectual rather than physical tasks. Robots still are limited in their ability to interact with the physical world: Scientists find it easier to program a computer to play advanced chess than to teach it how to find the pieces and move them from one square to another.
About 100,000 robots are now in use, mostly fixed in place and mostly specialized to particular tasks. A “universal” robot, able to move independently and perform various tasks, will need about 1 million times the power of a mid-1990’s home computer. Moravec sees this happening within the next several decades because of rapid advances in computer technology. Robots now are able to learn, rather than requiring preprogramming; they can generalize and plan ahead. Some can teach themselves more effectively than they can be programmed.
Moravec presented similar thoughts in his 1988 book Mind Children: The Future of Robot and Human Intelligence. The last ten years have seen tremendous...
(The entire section is 1993 words.)