Sightings: Turning an Old Leaf

Worth Noting

Recent reads that caught our attention.

Beyond Buzz: Thinking About Feelings

Logic will get you only so far when managing a PR crisis.

HR: You're Doing It Wrong - Is Your HR Leader a Glorified Administrative Assistant?

Too often, HR people are excited about the wrong things.

Workspace: "She's a Girl, But She Has Brains"

What we say when not talking explicitly about diversity says a lot about our thoughts on it.

Theory to Practice: Dangerous Digits

The perils of confusing measuring numbers with making them.


Forced Leisure

Where Will the Jobs Go?

Summer 2013

Where Will the Jobs Go?tcbrPDF normal

Technology may soon make obsolete factory workers, professors, nurses, and Teamsters.

By Jaron Lanier

Where Will the Jobs Go?

Jaron Lanier is a computer scientist, composer, visual artist, and author. Adapted from Who Owns the Future? (Simon & Schuster) ©2013

When it comes to the future of manufacturing jobs, the key question isn’t, “How much will be automated?” It’s how we’ll conceive of whatever can’t be automated at a given time. Even if there are new demands for people to perform new tasks in support of what we perceive as automation, we might apply antihuman values that define the new roles as not being “genuine work.” Maybe people will be expected to “share” instead. So the right question is, “How many jobs might be lost to automation if we think about automation the wrong way?”

The particular way in which we are digitizing economic and cultural activity will ultimately shrink the economy while concentrating wealth and power in new ways that are not sustainable. And that mistake is setting us up for avoidable traumas, as machines get much better in this century.

One of the strange, tragic aspects of our technological moment is that the most celebrated information gadgets, like our phones and tablets, are made by hand in gigantic factories, mostly in southern China, and largely by people who work insanely hard in worrisome environments. Looking at the latest advances in robotics and automated manufacturing, it’s hard not to wonder when the labors of these hordes of new potential Luddites might become suddenly obsolete.

In this case, even once the technology becomes available, I suspect politics will slow it down a little. It’s hard to imagine China deciding to throw much of its own population into unemployment. It is still a centrally planned society to a significant degree. It’s hard, even, to imagine one of China’s neighbors doing it. Would an aging Japan automate its factories to undercut China? Seems like a significant risk.

But somebody somewhere may find the motivation. Maybe a low-population but capital-rich Persian Gulf nation worried about the post-oil future will fund gigantic automated factories to undercut China in the production of consumer electronics. It might even happen in the United States, which has ever-fewer manufacturing jobs to protect anyway.

What would it look like to automate manufacturing? Well, the first word that comes to mind is temporary. And the reason is that the act of making manufacturing into a more automated technology would inherently move it a step closer to being a “software-mediated” technology. When a technology becomes software-mediated, the structure of the software becomes more important than any other particularity of the technology in determining who will win the power and the money when the technology is used. Making fabrication software-mediated turns out to be a step toward making the very notion of a factory, as we know it, obsolete.

A Factory in Every Home?

To see why, consider how automated manufacturing might advance. Automated milling machines and similar devices are already ubiquitous for shaping parts, such as forms for molds; robotic arms to assemble components are less common but still present in certain applications, such as assembling parts of large items like cars and big TVs. Detail work (like fitting touchscreens into the frame of a tablet) is still mostly done by hand, but that might change soon. At first, manufacturing robots will be expensive, and there will be plenty of well-paying jobs created to operate them, but eventually they will become cheap and the data to operate them might then be crowdsourced, sending manufacturing down the same road traveled by the recorded music industry.

Making fabrication software-mediated turns out to be a step toward making the very notion of a factory, as we know it, obsolete.

Consider 3D printing, which in a matter of months has graduated from academic theory to hobbyist dream to Staples inventory item and, in the form of home-printed guns that can fire, security threat. A 3D printer looks a little like a microwave oven; through the glass door, you can watch roaming robotic nozzles deposit various materials to form a product as if by magic. You download a design from the ’net, as if you were downloading a movie file, send it to your 3D printer, and come back after a while. There, before you, is a physical object, downloaded from afar. There are fledgling experiments with printers that realize physical products, including working electronic components. A chip is just a pattern deposited by something like a printing process to begin with; so is a flat display. In theory, it ought to be possible, in the not-so-distant future, to print out a working phone or tablet.

The key point: Once a 3D printer can be deployed in a factory, it might just as well be placed close to where the product will be used.

Being able to make things on the spot could remove a huge part of humanity’s carbon footprint: the transportation of goods. Instead of fleets of container ships bringing tchotchkes from China to our ports, we’ll print them out at home, or maybe at the neighborhood print shop.

What will be distributed instead will be the antecedent “goops.” These are the substances squirted out by the printer’s nozzles. Right now, there are about one hundred goops in use by 3D printers. For instance, a particular goop might harden into the kind of tough plastic found in car interiors. It is too early to say what goops will be in use in the future. Nor do we know how many different goops will be needed. Maybe a single supergoop would go a long way. Perhaps a suspension including graphene particles will be configurable into a variety of components such as nanotube digital circuits, battery layers, and tough carbon-fiber outer shells.

Will there be goops delivered by pipes to the home? Goop trucks that make rounds to refill printers once a week? Goop refill kits sold by Amazon and delivered by parcel? Little blimps that alight on your roof to refill your home printer? This we do not know. At any rate, a new infrastructure will be needed to get goops to printers. Expect goop to be as overpriced as ink for home photo printers is today.lanier3

The real magic might come about because of the transformation of recycling. Right now, when we throw something away, no information is packaged with that thing that described how it could best be disassembled into its constituents in order that they might be reused. This is a great inefficiency. We rely on human labor to very approximately assess what we toss away so that it can be recycled. This happens when we choose the right trash bin at the cafeteria, or when poor people pick over garbage dumps.

Once 3D printers commonly create objects, the nature of recycling will transform utterly. An object that had been printed will be remembered in the cloud. There will be “deprinters” that accept objects that are no longer wanted, like the previous year’s tablet. By referring to the original printing specification, always retrievable online, it will be possible to unravel the object back to its original goops with precision. Instead of melting it down, little nozzles with specialized solvents and cutting tools will separate each striation that originated from a different antecedent goop. The process will not be perfect, since the laws of thermodynamics cannot be revoked, but it will be hugely more efficient than what we do today.

Between the obsolescence of shipping and an extreme increase in recycling precision, 3D printing could create a massive explosion of convenience and fun and, at the same time, vastly reduce humanity’s carbon footprint and reliance on nonrenewable resources.

Doesn’t Someone Have to Make the Printers?

Of course, we don’t yet know about all the gotchas yet to come. But supposing that even some portions of the benefits appear, it certainly would be foolish to oppose this stream of progress. How could a liberal not like the smaller carbon footprint? How could a conservative not like the efficiency? And of course techies will be in love.

And yet the transformation will throw factory workers out of work in a massive wave. Will China be destabilized? As happened with the file-sharing of other things like music, the transformation of fabrication into a file-sharing phenomenon could happen very quickly.

Once a 3D printer can be deployed in a factory, it might just as well be placed close to where the product will be used.

When I explain this scenario, I often receive this response: “But someone still has to make the printers.” Somehow it’s hard to wrap our heads around a world in which the printers themselves are printed. You wouldn’t go buy a 3D printer at Wal-Mart. Your neighbor would print your first one for you. They’d spread “virally,” to use the usual metaphor.

Huge benefits on both a global and individual scale could appear, but coupled with a wave of supposed human obsolescence. I repeat that it’s only “supposed” obsolescence, because all those files that are shared to describe objects to print have to come from somewhere.

In a world of efficient 3D printing and recycling, we might experience much, much faster turnaround in our material culture than we are able to easily conceive of today. A guitarist might routinely print out a new guitar for every gig. Snobs might very well then decry that much of the design churn is stupid and pointless, just as critics might say the same about today’s social-network kinetics. But if people are interested in finding the latest stupid cool guitar to fabricate for the day, there will be a stupid cool guitar designer out there who ought to be paid.

The most radical change in daily life might be associated with fashion and clothing. A home device will be able to print out clothes based on Internet designs, but also based on your body. The device would scan your body in three dimensions, just as Microsoft’s Kinect input device does today. You’d see an outfit slinking about on your body before it exists. Everyone will be dressed exquisitely because every piece of clothing will be custom-fit.

Forget laundry. At the end of the day, you’ll pop dirty clothes into the top of the device for recycling. Never wear the same dress twice. (Though there will no doubt also be a countertrend in which vintage and handmade clothing becomes ever more revered. This is what happened with vinyl records after music became networked.) Today, “cool hunters” comb impoverished neighborhoods, sniffing out fashion trends. In the future, kids in those neighborhoods should earn wealth for their fashion trendsetting.lanier4

Napsterizing the Teamsters

Humans are terrible drivers. We kill each other in car accidents so frequently that the toll has become a more deadly problem than war or terrorism. It’s one of our biggest sources of death and pain.

Could self-driving cars do better? Definitely. Researchers around the world have begun showcasing cars that are quite effective at driving themselves, and results from experiments thus far indicate that it is unlikely robots will ever drive as badly as people.

The effects would be wide-ranging. For instance, stoplights would generally go away. Cars would simply know when there’s no other car coming, and no pedestrian, so they could just proceed through without stopping when there is no need. This would bring a huge gain in energy efficiency, since vehicles wouldn’t have to accelerate from a stop nearly as often. City driving would become almost as efficient as freeway driving.

If cars could coordinate with each other, traffic jams might become nearly extinct. Instead of people engaging in tiny ego-wars to merge between lanes on the freeway, causing huge backups going miles back, cars would anticipate mergers and merge cleanly, taking full advantage of the hypothetical bandwidth of the freeway.

Could self-driving cars do better? Definitely. It is unlikely robots will ever drive as badly as people.

True, there will be gotchas, just as with 3D printing, and we can’t yet know what they will be. When there is a screw-up, it could be a huge one. If a whole freeway of cars hit each other because of a snag, that would be a calamity on the order of a plane crash instead of an incident involving only a few people. That’s conceivable should there be many cars connected together virtually, moving rapidly under a connected software system.

What sort of economic impact will self-driving vehicles bring? It could be catastrophic. A giant portion of the global middle classes works behind a wheel; many entered middle-class life as a taxi driver or truck driver. It’s hard to imagine a world without commercial drivers. A traditional entry ramp into economic sustenance for fresh arrivals to big cities like New York would be gone; wave after wave of immigrants have driven New York taxis. And I’m trying to imagine the meeting when someone tries to explain to the Teamsters that nothing like their services will ever be needed again.

Both cabbies and truckers have managed to build up some legal heft over the years, so they’ll be able to delay the change—but not for long. For a while, there might be a compromise in which a Teamster or a cabbie sits there passively, along for the ride, perhaps to man a failsafe button. But young people won’t expect that to last and won’t seek it as a way of life. The world of work behind the wheel will drain away in a generation.lanier5

Degree vs. Dossier

The “creative classes,” including recording musicians, journalists, and photographers, have drawn much of the attention paid to vanishing economic dignity, not to mention the significantly larger number of people who supported these types of creators, like studio musicians and editors, and enjoyed “good jobs” (meaning with security and benefits).

The real question is whether the felling of creative-class careers was an anomaly or an early warning of what is to happen to immeasurably more middle-class jobs later in this century.

Why are we still bothering with higher education in the network age? We have Wikipedia and a world of other tools.

For instance, higher education could be Napsterized and vaporized in a matter of a few short years. In the world of the new kind of network wealth, towering student debt has become yet another destroyer of the middle classes.

Why are we still bothering with higher education in the network age? We have Wikipedia and a world of other tools. You can educate yourself without paying a university. All it takes is discipline. Tuition pays for making discipline a little more structured, getting some extended years of parental support in a place with a quad and beer, and certification. You also meet elite friends. There’s prestige in getting into a top school, whether you finish or not.

All of these benefits might be had less expensively in other ways, and that is becoming truer every day. The knowledge is no longer held in a dungeon. Anyone with a ’net connection can pretty much get any information that would be presented in a university. Undoubtedly some sort of social coercion site or fantasy game will take off online to help out with the discipline of self-education. As for the degree, the piece of paper, Internet statistics ought to be able to make mincemeat out of old-fashioned degree-earning in very short order. Why make do with a GPA when you can get a detailed dossier on your potential hire?

As for the years of parental support, it is turning out that in a Napsterized overall economy, more and more graduates stay with their parents well after college anyway. Why spend a ton of money supporting kids in college for four years when the same money could last longer to put them up somewhere cheaper?

Education Without Educators

I remember looking at images of all the bright young people in Egypt’s Tahrir Square, right after they had overthrown a dictator. Here was a forward-looking, young, savvy, and high-tech new generation. How would they get jobs? Shouldn’t a bunch of these young people be professors in Egyptian universities in ten years? Is the Internet going to make it easier or harder for them to get those jobs?

This is a pattern we’ll see over and over again when people interact with top network servers. You get an incredible bargain up front, like super-easy mortgages, insanely cheap retail items, or free online tools or music, but in the long term you also face reduced job prospects. In this case you get free online education up front but fewer academic jobs in the longer term.

We need to find a way to make education more available and make the career benefits of education more attainable.

Now, as levees break and austerity rules, suddenly contracts turn out not to be inviolable. This is what union members and copyright holders have learned. So where will this leave academics, as our century of digital networking proceeds? They’ll be caught clinging to the pomp of graduation ceremonies, to their employment contracts, and the tenure process, which has lasted for centuries. But all this will be under assault.

The problem won’t be the price of the buildings or the land of the campus. No, it’s always possible to raise millions of dollars to build a building, even as graduate students are paid so little that they take on lifetimes of debt just to make it through. Buildings are wealth, and wealth begets wealth. Graduate students are not.

How did anyone ever afford education? Society will not be able to afford the risk of the great debt load that students collectively take on. Austerity will force a contraction of government support of the academy, everywhere in the world at once.

Is it a coincidence that formal education is starting to become impossibly, cosmically expensive just at the moment that informal education is starting to become free? No, no coincidence. This is just another little fractal reflection of the big picture of the way we’ve designed network information systems. The two trends are a single trend.

I imagine that the academics from top technical schools will do fine. Honestly, there’s no way Silicon Valley would stand to see MIT fall. That wouldn’t be a danger anyway, because the top technical schools make money from technology. Stanford sometimes seems indistinguishable from a Silicon Valley company.

What about liberal-arts professors at state colleges? Some academics will hang on, but the prospects are grim. A decade or two from now, if nothing changes, the outlook will recall the present state of recorded music. In the case of that industry, making a pre-digital system efficient through the use of a digital network quickly shrank it economically to about a quarter of its size. It will shrink perhaps to about a tenth once people with old habits die off. This is not because of obsolescence. Music is not fading away like buggy whips, any more than the need for education will. Instead, wealth is becoming concentrated, since most of the real value, which still occurs out in the real world, on the ground, is reconceived to be off the books.

The lure of “free” will beckon. Get educated for free now! But don’t plan on a job as an educator.lanier6

The Robotic Bedpan

One of the bright spots in the future of middle-class employment is usually taken to be health care. Surely we’ll need millions of new nurses to care for the aging baby boomers. Caregivers will become a huge new middle-class population. If you want to think in terms of social mobility, this would also mean a huge transfer of wealth between generations that isn’t necessarily kept within families. It should be an example of the great wheel of middle-class aspiration turning anew in the United States.

But this prospect, like others, is unraveling. Just look at Japan.

The country faces one of the world’s most severe depopulation spirals. Around 2025 or 2030, Japan can expect a profound shortage of working-age people and a gigantic population of elderly people. Japan has traditionally not welcomed waves of non-Japanese immigrants. And it is at the cutting edge in robotics research.

Therefore, there is talk that robots will become advanced enough in time to take care of the elderly. This is plausible, from a technical point of view. Robots are already able to handle delicate tasks, like certain surgical subroutines, and are getting to be reliable enough to be a less risky choice than humans in some situations, such as driving vehicles.

There is talk that robots will become advanced enough in time to take care of the elderly.

Would a robot nurse be emotionally acceptable? Japanese culture seems to have anticipated the coming demographic crunch. Robots have been cute in Japan for decades. Trustworthy fictional robots, like Transformers and Tamagotchis, are a primary national cultural export. As with all waves of technological change, it is hard to predict when the inevitable glitches and gotchas will be smoothed out. In this case, though, the motivation is so intense that I expect robots in Japanese nursing homes by 2020 and in widespread use by 2025.

Sans robots, one would expect waves of immigrants to go to American nursing schools in the next decade to prepare to take care of America’s own age wave. Their children would be raised by parents who practiced a profession, and would tend to become professional themselves. Thus a whole new generation of customers for colleges and a new wave of middle-class families would make their way, continuing the American pattern.

But those imported robots will be awfully tempting. If you spend any time in eldercare facilities such as nursing homes, a few things become apparent. First, there is no way for even the most professional and attentive staffs to help everyone as fast as would be ideal. It’s inconceivable to have immediately available 24/7 help for every discomfort that comes up. Second, eldercare is unbelievably hard and uncomfortable work, if it’s done well. It’s very hard for even the best facility to make absolutely sure that every member of the staff is always doing the best possible job. The elderly make easy victims, like children. Petty thefts and taunts are not uncommon.

It’s not that robots will necessarily be immediately cheaper than human staff. There might be significant expenses associated with the goops needed to print them, if they’re printed, or with manufacturing and maintaining them if they are not. But the expenses will be more predictable, and that will make all the difference.

Hiring a human nurse will mean paying for that person’s health insurance, and taking on unpredictable legal liabilities for the mistakes that person might make, like leaving a floor wet. Both of these drags on the ledger will be amplified by network effects, just as has happened with mortgage risks.

Humans will always do those jobs that a robot can’t do, but the tasks might be conceived as being low-skilled.

Insurance companies will use computers to weasel out of liability and to extract ever-larger payments. The whole world’s lawyers will be circling online. The liability side of having an employee will be copied and amplified over a network, just like a pirated music file or a securitized mortgage. It will eventually become less risky to choose a robot. When you turn action into software, then no one gets blamed for what happens.

Humans will always do those jobs that a robot can’t do, but the tasks might be conceived as being low-skilled. It might turn out that robots can give massages but can’t answer the door. Maybe robots will be good at catching patients who fake the ingestion of medicines but ineffective at soothing patients so that they’ll take them voluntarily.

The key reason to avoid acknowledging that there’s real skill in doing what robots can’t do—and hiring people for real jobs—will not be to keep the immediate expenses low but, rather, to reduce the amplified liabilities of the network age. So there will be plenty of dead-end jobs without security or benefits. This will be despite the fact that the humans in the caregiving loop might be absolutely essential to the well-being of those being cared for.

The latest waves of high-tech innovation have not created jobs like the old ones did. Iconic new ventures like Facebook employ vastly fewer people than big older companies like, say, General Motors. Put another way, the new workforce schemes that grant ultimate power to networked computers channel much of the productivity of ordinary people into an informal economy of barter and reputation, while concentrating the extracted old-fashioned wealth for themselves. All activity that takes place over digital networks becomes subject to arbitrage, in the sense that risk is routed to whoever suffers lesser computation resources.

The universal advice of our times is that people who want to do well, as information technology advances, will need to double down on their technical educations and learn to be entrepreneurial and adaptable. But without considering technology’s impact on middle-class employment, there won’t be enough positions working with networks and computers to sustain a society.


The Conference Board

From the Archives

The Conference Board Review is the quarterly magazine of The Conference Board, the world's preeminent business membership and research organization. Founded in 1976, TCB Review is a magazine of ideas and opinion that raises tough questions about leading-edge issues at the intersection of business and society.