• U.S.

The Robot Revolution

30 minute read
Otto Friedrich

COVER STORY

For good or ill, it is already transforming the way the world works

The new robots do not really look like Frankenstein’s monster, or like Artoo Deetoo in Star Wars, but rather like a row of giant birds.

They poke their 9-ft.-long, rubber-sheathed necks toward the row of automobile frames. From their beaks, a blinding shower of sparks streams forth. The escape of compressed air creates a loud hissing sound. This is Chrysler’s sprawling 145-acre Jefferson plant in East Detroit, where the trouble-ridden firm is building the new K-cars—the Plymouth Reliant and Dodge Aries—that it hopes will save its future. Once 200 welders with their masks and welding guns used to work on such an assembly line. Here there are no welders in sight; there are only 50 robots craning forward, spitting sparks. They work two shifts, and the assembly line’s output has increased by almost 20% since the robots arrived earlier this year.

In a plant outside Turin, the Italian firm of Digital Electronic Automation is trying out its first new Pragma A-3000. The $110,000 robot, which has just been licensed by General Electric, is assembling a compressor valve unit from twelve separate parts. Its two arms can do totally different jobs at once. When it picks up a slightly defective gasket in its gray steel claw, it immediately senses something wrong, flicks the gasket to one side and picks up another. The Pragma produces 320 units an hour, without mistakes, and it can labor tirelessly for 24 hours a day. That makes it roughly the equivalent of ten human workers. Furthermore, it can easily be reprogrammed to assemble TV sets or electric motors or, theoretically, just about anything.

Near Golden, Colo., at the Department of Energy’s Rocky Flats plant, a technician pushes a red button marked REQUEST TRANSFER. Behind a 10-in.-thick concrete wall, a pair of claws reaches out to grasp a stainless steel container filled with pink powder, then lifts it into a furnace where it is baked at 950° F until it turns into a nondescript gray button three inches in diameter. Such a button could be worth $100,000, for the job of this robot, which goes into regular operation in a few months, is transporting reprocessed plutonium, one of the most toxic substances known to man. Until now, this dangerous task has been done by men in elaborate space suits. The robot, which knows neither weariness nor boredom, also knows nothing of danger.

The robot, a dream as old as man’s yearning to avoid doing his chores (see box), is finally emerging from the pages of science fiction and beginning to transform the way the world works. What this amounts to is nothing less than a robot revolution. It promises to revive decaying industries and give smaller firms all the benefits of mass production. Ultimately, it may also transform the way society itself is organized and the way it assesses its values. These steel-collar workers already paint cars, assemble refrigerators, drill aircraft wings, mine coal and, for that matter, wash windows; newer robots now on the drawing boards will soon be spraying crops with pesticides, digging up minerals deep under the oceans and repairing satellites in outer space. Not too far off, experts predict, is that landmark day when robots will begin designing and then building other robots. “The human race,” according to James S. Albus, head of the robotics research laboratory at the National Bureau of Standards in Gaithersburg, Md., “is now poised on the brink of a new industrial revolution that will at least equal, if not far exceed, the first Industrial Revolution in its impact on mankind.”

That first revolution, which began two centuries ago, created the technology of modern life, but at a high cost in hardship and hunger. Some experts see analogous dangers in the robot revolution. If robots can do men’s work faster, better and more cheaply, then what will men do? They will be retrained for other things, the robotmakers answer. But by whom, and for what? Almost 20 years ago, Kurt Vonnegut’s Player Piano portrayed a future society in which the elite few run the machines while the unemployable majority subsists on handouts in resentful idleness. “It’s an enormous problem,” concedes Luigi Lazzaroni, president of the Italian firm that makes the Pragma robot. “Many will have to learn how to work differently. The schools, the industrial firms and the government must cooperate to ensure that the workers are able to fit the requirements of industry.”

In the U.S., the robot revolution originates in American industry’s most fundamental problem: the stagnation in productivity. From 1947 to 1965, U.S. productivity increased by 3.4% a year, but the growth rate dipped to 2.3% in the following decade, then dropped to below 1% in the late 1970s and down to —.9% last year. (Japan’s productivity growth, by contrast, has been climbing at an average annual rate of about 7.3%.) Now that economic planners are trying to work out methods of “reindustrializing” the U.S., they can see in the robot a major answer to those productivity declines.

Not only can the robot work three shifts a day, but it takes no coffee breaks, does not call in sick on Mondays, does not become bored, does not take vacations or qualify for pensions—and does not leave Coca-Cola cans rattling around inside the products it has helped assemble. Its “up time” on the job averages around 95% (the figure for the average blue-collar worker is about 75%). In addition to its Horatio Alger work habits, it is immune to government and union regulations on heat, fumes, noise, radiation and other safety hazards. The robot has no affections or passions. If you prick it, it does not bleed. If you poison it, it does not die.

Two key developments have brought the industrial robot to life.

One was technological, the development in the mid-’60s of the microprocessor, a computer so small that it can be fitted onto a silicon chip no bigger than a pea. As the computer shrank in size and cost, it suddenly became practical as the brains to run a robot. The second development was wage inflation. Two decades ago, a typical assembly-line robot cost about $25,000; that, plus all operating costs over its eight-year lifetime, amounted to about $4.20 an hour, slightly more than the average factory worker’s wages and fringe benefits. Today that typical robot costs $40,000 (they range from $7,500 to $150,000), and it can still be paid for and operated at $4.80 an hour; the worker often costs $15 to $20. That is the formula for a gold rush.

The robot revolution is just beginning, but it is already moving fast. Scarcely a decade has passed since General Motors became the first major industrial robot buyer by ordering more than 50 welders. Today GM has 270 robots, and there are more than 3,000 at work throughout the U.S. The biggest manufacturer, Unimation Inc., of Danbury, Conn., was founded in 1959 and cost its parent company, Condec, at least $12 million before making its first profit in 1975. It now produces 40 Unimate and 15 Puma robots a month, and will have estimated sales this calendar year of $42 million. Its chief competitor: Cincinnati Milacron, which makes the sophisticated T3 robot and expects 1980 sales of $32 million. It will soon open a new plant in Greenwood, S.C. Sprouting up are newcomers like Automatix Inc., of Burlington, Mass., which was founded last year with $6 million from, among others, Harvard and M.I.T. Giants like IBM and Texas Instruments are weighing the advantages of getting in on the prospective bonanza. Overall, the fledgling U.S. robot industry is producing about 1,500 units per year and is projecting sales of $90 million this year. Wall Street analysts predict a growth on the order of 35% a year throughout the 1980s. That gives the industry a sales potential of more than $2 billion by 1990. Boosters talk of $4 billion.

Spurring U.S. manufacturers is the fact that foreign competitors are already ahead in many ways and fighting to dominate the future. Chief among them are the Japanese, who imported their first Unimates in 1967 and now operate most of the robots in the world (about 10,000, compared with the 3,000 in the U.S. and about the same number in Western Europe-). They are also outproducing the U.S. in robots at a rate of at least 5 to 1. Like many troubled U.S. executives, General Electric’s Julius Mirabal recalls going to Japan in 1976 to compare production techniques. He found robots everywhere, including one cluster that had reduced the work force in a vacuum-cleaner plant from several hundred men to eight. “Unless we start doing something to increase U.S. productivity, the United States will be out of business as a country,” says Mirabal, who returned from Japan to find that GE was using only ten robots; today it has 111. The auto industry now buys about 40% of industrial robots, both in the U.S. and worldwide, but electrical firms have also become major users.

The Japanese, meanwhile, are resolutely pressing forward. In January Fujitsu Fanuc will open a new $38 million plant in which robots will work 24 hours a day to produce more robots (100 a month). “The danger in letting Japan get so far ahead,” says Paul Gosset, who helped develop robots for France’s Renault, “is that they may end up being the ones who make the modules and parts that go into everyone else’s robots.”

Webster’s definition of a robot begins by describing it as “a machine in the form of a human being that performs the mechanical functions of a human being.” Today’s robotmakers, however, are devoting very little thought to creating anything that looks or acts human. It is perfectly possible to design a robot that walks on artificial legs or speaks fluent English, but it is much cheaper and more efficient to keep the robot standing in one place and to speak to it in the soothing language of algorithms. Says David Nitzan of SRI International: “We’re creating an image of a robot in the way Picasso redistributed the features of a human face.”

A robot’s basic function is not to look or behave like a human being but to do a human’s work, and for that it needs mainly a guiding brain (the computer) and an arm with claws for fingers. The computer is simply plugged into an electric outlet; cables run from the computer along the robot’s arm and transmit instructions in the form of electric impulses to the claw; for heavy work, robots use hydraulic pressure. The Robot Institute of America, an industrial trade group, therefore offers a contemporary, if somewhat prolix, definition of a robot: “A reprogrammable, multifunctional manipulator designed to move material, parts, tools or specialized devices, through variable programmed motions for the performance of a variety of tasks.”

Reprogrammable and multifunctional are the key words. Factories have long used automatic machines (like bottle cappers) to mass-produce goods, but these devices could only perform one task at a time. New work routines required new machinery or extensive retooling. The industrial robots now being installed have control and memory systems, often in the form of minicomputers. These enable the robots to be programmed to carry out a number of work routines and, when necessary, to be reprogrammed to carry out even more.

The fact that the robot’s instructions can be changed is critically important to its industrial use. A standard assembly line must produce a large amount (about 1,000 units a day in the auto industry) to operate economically, and it takes months to alter or renovate its component machines; a robot can be reprogrammed for a new task in a few minutes. Furthermore, at least 60% of U.S. manufacturing is done in batches too small for assembly lines. Robots can do many of those jobs, and it is estimated that they can reduce costs in small-lot manufacturing by 80% to 90%.

Now that robots have proved efficient and economical, the main effort is to create “smart” robots and thus give them an ability to make decisions. To become smarter, robots are learning to “see” and “touch,” and report to their computer brains what their new senses tell them. To see means to decipher what appears before a TV camera; to touch means to measure not only the size and shape but the temperature, softness or vibration of the object grasped by the claw. Robots can also hear, and could presumably be taught to taste and smell, but these would be mainly indulgences, not necessary to their work ethos. On the other hand, robots are now being outfitted with senses that no human being has: the perception of infrared light and ultrasonic sound.

General Motors has developed a system called Consight that enables a robot equipped with an electronic camera to look at scattered parts on a conveyor, pick them up and transfer them in a specific sequence to another work area. It thus makes rudimentary judgments on which parts to pick up, but it is still too slow for an industrial assembly line. At a well-attended robot exhibition last month in Dearborn, Mich., one of the star attractions was a similar vision system developed by a brand-new company, Machine Intelligence Corp. of Mountain View, Calif. This firm was founded in 1978 by Charles Rosen, 63, a tall, tousledhaired veteran of 21 years at SRI. Says he of his new vision system: “It’s still only the beginning. We’re at a stage akin to when the first lathe did a reasonable job on a hunk of metal. But machine vision has as important a role in automated assembly as human vision has for assembly by humans.”

Rosen’s system consists of a black-and-white TV camera that scans objects against a brightly lighted background, then transmits to a computer the hundreds of dots (or pixels) that form the TV image. The computer transforms these dots into binary code and compares what it sees with previously recorded descriptions of various objects. It compares features like perimeter and area, enabling it to recognize and choose among nine different objects. “Ten years from now,” says Rosen, “this will be a dodo.”

At the Lockheed Missiles and Space Co., engineers are already pushing one step further into a technique called “gray-imaging.” Similar to Rosen’s system but more elaborate, the Lockheed method uses a camera image that contains 100,000 different dots, each graded from 0 for pure white to 255 for pure black. The different shades of gray give the robot a much clearer three-dimensional view of what it is confronting.

Lockheed’s project, which started with an Army contract to search for means of spotting defective artillery shells, is only one of many robot efforts sponsored by military and space programs. The most spectacular, of course, is the Voyager 1 robot, which traveled 1.3 billion miles to Saturn. Almost equally impressive is the Mars Rover being built by CalTech’s Jet Propulsion Laboratory in Pasadena, Calif., which will be able to wheel itself about on the rugged planet, look at rocks with its TV eyes and dig up samples with its shovel. Engineers at the Marshall Space Flight Center in Huntsville, Ala., now are working on a robot that will be able to take off from the space shuttle, reach an ailing satellite in orbit and repair it. The Naval Research Laboratory in Washington, D.C., similarly, is building a robot that can be sent out aboard an unmanned submarine to find and repair crippled vessels undersea. Robots are already at work in the manufacture of tanks, aircraft, guns and ammunition.

There are some experts, though, who believe that sight is much less important than touch, either undersea or on the assembly line. “I can’t afford to let the robot arm wait while the camera does all the things it needs to do,” says GE’s Mirabal, who says he has looked at 20 vision systems and found none that is economical. “Touch is going to be very important, because all the robot needs is to know that something is happening or not happening. Just one piece of information that can be analyzed quickly.” While most of the touch systems are developing a robot claw’s ability to measure objects, some are more elaborate. The Lord Corp. of Erie, Pa., hopes to market within five years a “hand” made out of spongy material with a grid of many sensitive wires embedded in it to achieve a true sense of touch.

Robots sometimes seem remarkably stupid to the engineers trying to educate them. A robot can cope with complex mathematical formulas, of course, but when it sees something through its TV camera, it has a hard time translating the two-dimensional image into three-dimensional reality. A robot instructed to look for a triangular object will waste valuable time fingering cubes and cylinders before rejecting them. And when a component burned out in a robot at the University of

Florida, the machine suddenly hit itself so hard that it sheared off its arm. Says John Dixon, a computer scientist working on the Navy’s underwater explorer: “We humans have been manipulating things ever since we were children, so we’re extremely good at it. But if you analyze everything that’s going on when you do a simple thing like picking up an object, it’s really very complicated.” Adds David Grossman, I.B.M.’s manager of automation research: “It’s like trying to write down how to tie a shoe.”

At M.I.T., great labor has gone into creating a robot that can watch someone constructing an arrangement of toy blocks and then duplicate that arrangement. Engineers at Japan’s Waseda University built a robot seven years ago that could see and hear and carry out spoken instructions, but, says Ichiro Kato, chairman of the graduate school of science and engineering, “it had the mentality of a child 1½ years old.” Kato’s lab is now building a more advanced model. Says Kato: “It will probably have the mentality of a five-year-old.”

Still, the main function of an industrial robot is not to think but to work, and there are many jobs that a sufficiently muscular and adroit five-year-old could do admirably. At Pratt & Whitney’s automated casting factory in Middletown, Conn., ten of Unimation’s Unimate 2000s are building ceramic molds for the manufacture of engine turbine blades. The company expects the new molds to help increase production from 50,000 to 90,000 blades a year. No less important, the robot-made molds are so much more uniform that their blades last twice as long as blades molded by humans.

At the General Dynamics plant in Fort Worth, one of Cincinnati Milacron’s T-3 robots makes sheet-metal parts for the F-16 fighter. The T-3 selects bits from a tool rack, drills a set of holes to a .005-in. tolerance and machines the perimeters of 250 types of parts. A man doing the same job can produce six parts per shift, with a 10% rejection rate. The robot makes 24 to 30 parts, with zero rejections. The machine costs over $60,000 and has saved $93,000 in its first year.

In a noisy inferno at Westinghouse’s lamp factory in Bloomfield, N.J., a Unimate 2015G robot performs a process called “swaging.” This is somewhat like making spaghetti, but it is done with 21-in. rods of yellow tungsten, destined to become light-bulb filaments. The robot lifts them off a conveyor belt and sticks them into a blazing furnace (3,200° F), then into a swaging machine that stretches the rods until they have grown to 37 in. in length and shrunk to exactly .467 in. in diameter. Three workers, each of whom cost the company $20,000 per year, used to do this very unpleasant labor with increasingly uneven results during their eight-hour shifts. The robot does it flawlessly for 16 to 24 hours a day. It will pay for itself in 2½ years.

At the Chesebrough-Pond’s thermometer plant at 98.6 Faichney Drive in Watertown, N.Y., a Unimation Mark II is in charge of the delicate task of removing any air bubbles that may remain in the mercury inside a thermometer. Established in an isolated room, because of the increased awareness of the dangers of mercury poisoning, the robot takes a boxful of thermometers and lowers it into a tank of hot water (100° F to 145° F), then into a tank of cold water (40° F), then into a centrifuge that squeezes out even the tiniest bubbles. Working with two dozen different boxes, it performs its ritual three times on each box in the course of a 394-step program that takes 7½ min. A simple routine, but it used to occupy 13 employees, and now only one is necessary. Says Plant Manager M. James Dawes: “I tell our people we’ve got to become more productive by being smarter, not by working harder.”

This sense of the robot as a helper rather than a menace is widespread among factory hands. Though robots are highly vulnerable to sabotage, there has been no trace of the Luddite violence that threatened the first labor-saving machines of the Industrial Revolution. On the contrary, working with a robot seems to confer status. And, while the machine usually looks less like a man than like a lobster, its human partners often seem unable to resist giving it a name and even lavishing on it a certain metallic affection. When one machine known as “Clyde the Claw” broke down at a Ford stamping plant in Chicago, its human partners gave it a get-well party. Chauvinism being what it is, most factory workers unthinkingly refer to a robot as “he,” but at one plant in Japan the clanking automata have each been given the name of a female movie star.

The willingness of the robot to do the dirty work, like some mechanized Turkish Gastarbeiter, has muted alarms about the loss of jobs and has kept the labor unions mostly at bay. Welding cars and spraying paint are stupefying jobs, and, besides, they are ideally done at temperatures hotter than a worker can stand. “In the next five years,” says Anthony Massaro, Westinghouse’s chief of robotics technology, “we’re going to lose 25,000 people in manufacturing due to attrition, and there’s no way to replace them all. People joining the labor force these days don’t want the dirty jobs.”

Robert Cannon, president of an electrical workers local that represents many Westinghouse workers in New Jersey, accepts that reasoning. “Frankly, I welcome it,” he says. “If we can bring in a robot here to do, say, the painting that a man does for $7, then we can move him to another job at $7.50 an hour. We say, ‘Train our people for the skilled jobs that are in today’s market.’ ”

For the present, that is what is happening, and it can continue as long as corporations do not make the shift to robots faster than the natural rate of worker attrition, which now runs as high as 15% in the metalworking plants that are ripe for robotization. (One reason why Japan has been able to shift so extensively to robots is that Japanese corporations have a tradition of caring for their employees for life.) But as the robots take over more and more jobs—and they can do the more pleasant and interesting tasks as well as the dull and dirty ones—the unions’ acquiescence may change. The U.S. unemployment rate is already 7.6%, after all, and retraining programs have so far had little effect on it.

“Ultimately,” predicts Harley Shaiken, a former Detroit assembly-line machinist who now works as an industrial consultant at M.I.T., “retraining will not be possible, because there will be no jobs for workers to be retrained for.” That is probably an exaggeration, but Charles Cook, president of the United Auto Workers Local 7, which represents K-car workers at Chrysler’s Jefferson plant, is equally suspicious. Says he: “Our workers are not worried now about robots taking their jobs, but once the company gets more of those goddam things working, we’ll have problems.”

Already there are a few rumblings. Says Russ Cook, U.A.W. district committeeman at GM’s Buick plant in Flint: “If we don’t get smarter and start combatting the machines, we will be cannibalizing ourselves and competing against one another for jobs.” Adds Larry Jones, a Chrysler metal-shop worker: “They say they are only going to put robots on boring jobs. But in an auto plant, all the jobs are boring jobs.”

Aside from the specific problem of lost jobs, Shaiken warns of more intangible difficulties. “The use of robots has social costs that are not being addressed by anyone in the U.S. today,” he says. “By designing a production process that minimizes human participation, you freeze out the worker’s control and you freeze out his initiative. We often overlook the impact of robots on the jobs that remain. Today, if a worker assembling components has a daily quota of 100 units to fill, he can, for example, work flat out and assemble 60 in the first half of a shift, leaving only 40 for a relatively unpressured second half. But when he is slotted between centrally programmed robots that dictate the pace, he becomes a mere cog in the machine. These things matter.”

Leaders in the robot industry claim that the main resistance to their inventions comes not from union labor but from management. “We are thrusting ourselves into the manufacturing area, and it’s a very conservative place,” says Joseph Engelberger, the ebullient president of Unimation. Top executives turn for advice to their technical managers, and these are naturally cautious. “Plant supervisors get worried because they don’t understand robots,” says Neale Clapp, a robotics expert for the management consultant firm of Block Petrella Associates in Plainfield, N.J. “They feel their authority is undermined.” G.E., for one, commissioned a psychologist to study the effect of the introduction of robots on workers, and it found the greatest anxiety among foremen. Says James Clark, operations manager of the Westinghouse Elevator Co.: “The fear is: ‘What do I know about this? What will I be supervising? Will I be killed if it doesn’t work?’ ”

To all this, the robot backers offer two answers. One uses the hard language of survival. “If we don’t go to robots,” says an expert at Carnegie-Mellon, “we’ll just continue to lose to Japan and West Germany. Our economy won’t grow, and there won’t be any new jobs. New jobs have always come from new technology.” The other answer is a gentler prophecy of benefits to come.

“It’s my fervent belief,” says Engelberger, “that any increase in productivity is always good. The problem is to decide what to do with the blessings. Do we want to have a shorter work week? That’s one of the possibilities. Would we like clean air and water?

Three percent of the G.N.P. will give us a clean environment. The point is to separate out the problems, not just say, ‘Gee whiz, people are going to lose their jobs.’ So far, in any case, we’ve created a hell of a lot more jobs than we’ve displaced.” Adds Edward Fredkin, professor of computer science at M.I.T.: “We are creating what is going to be an immense new industry, perhaps as big as the auto industry.”

Not everyone, of course, is so euphoric about the coming robot age. Brian Carlisle, Unimation’s general manager for West Coast research, warns that “we’re a long way from a robot that can assemble a carburetor.” Nor are robots a panacea for all the ills that industry is heir to. The most automated factory of its time was the Lordstown plant that GM designed to produce the unsuccessful Vega, evidence that productivity is not worth much if the product is hard to sell. As the robotmakers look ahead, though, they see a promised land. It is a land in which the factory computers guide the original design of a product and then translate all instructions for the robots that provide the muscle on the automated assembly line. Furthermore, increased production should keep driving down . costs, both of the robots themselves (to an estimated $10,000 each) and of everything they produce. Says one forecast by the American Society of Manufacturing Engineers and the University of Michigan:

¶ By 1982, 5% of all assembly systems will use robotic technology.

¶ By 1985, 20% of the labor in the final assembly of autos will be replaced by automation. In the same year, “scene analysis” will provide enough feedback for robots to select parts scrambled in a bin.

¶ By 1987,15% of all assembly systems will use robot technology.

¶ By 1988, 50% of the labor in small-component assembly will be replaced by automation.

¶ By 1990, the development of sensory techniques will enable robots to approximate human capability in assembly.

While all these prospects are primarily industrial, U.S. planners are fully aware of the implications that a productivity revolution will have on a nation’s global power. Says the Bureau of Standards’ Albus: “Any country that develops the capacity to run its factories around the clock seven days per week with only a few human workers will have a tremendous advantage both economically and militarily. If nothing else, this capability would allow military weapons to be produced in virtually unlimited quantities at extremely low costs. But even assuming that such plants were never used for military production, the country that possessed such a large surplus of efficient production facilities could easily dominate the world economically.”

Outside the factory and the lab, the work that needs to be done in this world is almost without limits, and so is the robot’s potential ability to do it. In the field of farming and food processing, for example, Unimation has been asked to design a robot that can pluck chickens. Australian technicians are already testing robots to shear sheep. One machine first stuns the animal with an electric shock, then closes in with its shears. Clipping the back and sides is not too hard, but the technicians still report “significant difficulties” in finishing up the neck and head. In Japan, Mitsubishi has devised a robot that can visually distinguish different species and sizes of fish in a catch, then separate them into various bins with its mechanical arm. The company is developing similar robots to process fruits and vegetables.

And there is more. Care for the afflicted? Quadriplegics may some day use spoken commands to order robot servants to do their bidding. Other designers are working on a robot that could gently lift up a bedridden patient, while a nurse changes his sheets, and tuck him back into bed. M.I.T. Computer Scientist Marvin Minsky visualizes a day, about 20 or 25 years from now, when a surgeon will be able to slip on a pair of special gloves connected by remote control to a pair of mechanical hands that can perform surgery for him in a hospital hundreds of miles away. Fighting crime? The Advanced Robotics Corp. is advertising a mechanical sentinel that can speed to the site of any breakin, sternly ask an intruder, “What are you doing here?” and temporarily blind him with its spotlight while its siren calls for help.

But can robots wash the dishes? Of course, say the engineers. Unimation’s Engelberger, for that matter, is outfitting his office with a robot that will make and serve coffee to his guests. Fredkin of M.I.T. visualizes the household robot as a creature that could not only do all the chores but also chase away burglars, “preferably by crouching in a dark corner and growling like a large dog.” But does the ordinary homeowner want to pay $50,000 to get the kitchen sink cleaned up? “Actually, homes are a complicated environment for robots,” says one expert in Washington. “You’d have to build a robot that is 100 times more complicated than today’s industrial robots for one-tenth the cost.”

These are, furthermore, the service jobs that are supposed to be reserved in the future for the factory workers retrained out of the robot-run assembly lines. But wherever there is drudgery, the robot stands ready to move in. Says Westinghouse’s Clark: “If a robot can do the job, a man shouldn’t be doing it anyway.”

This idea that man is destined for higher things than work—not necessarily a realistic idea or even a meritorious one—provides the green light at the end of the pier. Says Albus: “The robot revolution will free human beings from the pressures of urbanization and allow them to choose their own life-styles from a much wider variety of possibilities.”

British Agriculture Minister Peter Walker suggests an even more heady vision: “Uniquely in history, we have the circumstances in which we can create Athens without the slaves.”

Such speculations raise com plex questions of social organization, starting with the matter of the wages paid for work done. If more and more work is done by robots, more and more people will eventually be living on some form of subsidy. Whether this takes the prestigious form of, say, long-term free education or the disdained form of welfare payments is a problem computers are already puzzling over.

Traditionally, earned wages define a person’s worth, in his own eyes as well as in those of his neighbors. But there is no law that this must be so.

The millionaire, the soldier, the vagabond and the poet all have other ways of judging their value. Says Science-Fiction Soothsayer Isaac Asimov: “Robots will leave to human beings the tasks that are intrinsically human, such as sports, entertainment, scientific research.” The value of many things will eventually change, since every thing made by machine will come to seem commonplace and everything made by hand, even a pair of knitted socks, will acquire the quality of rarity.

To guide all these potential changes will require a degree of ethical judgment and social organization that humanity has rarely shown any sign of possessing. Just as the computer itself derives, however, from the simple proposition that all mathematical logic can be reduced to various combinations of zero and one, these revolutionary upheavals in human society are clearly vis ible in the distance. Indeed, they can be seen already in the birdlike contraptions that poke their fiery beaks into the un finished steel frames at the Jefferson plant in East Detroit.

By Otto Friedrich. Reported by Christopher Redman/Detroit and Janice C. Simpson/New York

-West Germany has the largest number, 850. Sweden has 600, the most per capita in the world, and Sweden’s ASEA is the world’s third largest manufacturer of robots (after Unimation and Kawasaki). Italy has 500, France 200, the Soviet Union 25.

More Must-Reads from TIME

Contact us at letters@time.com