Joe Atikian is the author of Industrial Shift: The Structure of the New World Economy.
The robots are not coming for our jobs. In 1915, the industrialist Lloyd Raymond Smith asked, “Can automobile frames be built without men?” The same question has been asked about entire industries. It never happened, even though automation has appeared in every workplace. And it didn’t merely appear. Factory labour got intensively more mechanized from the moment factories were born. The surprise was that factory employment kept growing for another 65 years, including the early computer era.
Computer technology intensified automation, but it also spawned a whole new industry that almost nobody saw coming. It enabled much of the automation that finally started reducing U.S. factory employment in 1980, but even so, it’s not as one-sided as you might think. Since computers became a mainstream business tool, a million more IT-related jobs were created than jobs lost in manufacturing. And on the whole, these jobs in programming or computer sciences pay more. Median income for this job category now more than doubles the earnings from those good old factory jobs. This is technological progress, the largely unplanned creative destruction that Viennese economist Joseph Schumpeter described in the 1940s.
Advances in technology don’t guarantee that everyone will come out okay. When manufacturing work declines, current workers don’t easily move into IT jobs. While half of displaced workers find new work at equal or higher pay, perhaps in the new robotics field, a few will move into other kinds of work at lower pay, and then lose their jobs altogether.
So who gets the new IT jobs that replace older technology? Sometimes it’s the displaced workers, but for the most part it’s the next generation. Virtually no kids or young adults want to work in a factory or even know what a factory is. They don’t tend to make or fix things with their hands, much as a 1960s factory worker deliberately knew nothing about growing wheat. Instead, today’s youngsters are computer coders. They want to write apps for smartphones and dream of making a fortune in a startup.
Now enter artificial intelligence or AI. On one hand, computers are smart enough to assist in surgery. On the other hand, an artificial genius feeds you repeated ad messages asking if you’d like to buy the same item that you just bought last week. Even so, AI is clearly the next phase of workplace computer-robot automation, and deserves attention.
How can we estimate the future employment impact of robots combined with AI? One way is look back at recent impacts from related technology. Grocers have been adding self-checkout machines since the 1980s, and their numbers will soon rival bank machines. Yet cashiers remain one of the three largest stable occupations. We now have millions of ATMs, yet we still have nearly as many bank tellers as ever. The net employment result of computerized automation has often been near zero.
In a related shift, computers brought the proliferation of do-it-yourself tasks such as banking, travel bookings, classified ads, licence applications and tax preparation. We face a huge new burden online and in stores, yet traditional retail and service jobs persist. The artificially intelligent robots have already begun their impact and yet the human jobs are still there. So, the recent past shows us how computerized robot technology shifts tasks around without massively destroying jobs. Let’s use this experience as a tool to look forward.
Obvious is not always what it seems. We couldn’t have expected tellers to survive with ATMs, so now it’s fair to ask if other examples might confirm a similar positive trend. The airplane pilot must be the perfect example of an occupation under technological threat.
Air carriers would love to get rid of pilots, who comprise one of their biggest operating costs, and auto-pilot technology has been around for decades. It seems like a natural fit for modern AI and robotics, but the so-called “last mile problem” still precludes autonomous commercial jets. These jets are nearly 100-per-cent automated, but they still need a real pilot to taxi, takeoff, land and even to fly in rough conditions. Automation again proves to be an adjunct to human labour.
Truck drivers also appear imminently vulnerable to automation in long-haul transport. The threat looms large especially because driving is the last refuge of low- and mid-skilled occupations for American men (just as the checkout aisle seemed to be for women). Autonomous truck testing is under way, and remote control appears more realistic. Nevertheless, as with airplanes, the technology performs worst where it matters most, in the last mile. Self-driving trucks cannot back down city streets into a store’s driveway, or against a crowded neighbourhood sidewalk, so an on-board or remote attendant would still be employed.
The last-mile problem will virtually guarantee some form of trucking jobs forever, even if some industry sectors decline. And as with decades of experience with factory-computer employment outcomes, new sectors will arise. Smartphone apps won’t be the last invention in history.
Equally important is that none of this industrial history, from car frames to AI, was the result of a planned progression. The only kind of plan we need is one to assist the few displaced workers whose livelihoods get irretrievably sidelined by the technological developments that help everyone else progress.