Bring on the Bots
The antidote to the Great Resignation and a plunging birthrate is the robot revolution—which will not mean the end of work.
Elon Musk, never one for understatement, recently said that “one of the biggest risks to civilization” is that there “are not enough people” because the “foundation of the economy is labor.” He was reacting both to long-term trends and to the recent lockdown-induced collapse in America’s birthrate. Now, with the Great Resignation—working-age Americans voluntarily quitting jobs in huge numbers—in full swing, the challenges of a shrinking workforce have accelerated into our present.
Musk’s solution? Encourage people to have children (amen). And deploy robots. He’s right on both counts. But the robots can make a bigger difference, and faster, not only boosting productivity but also creating new jobs for humans.
Last year saw more than 1 million more baby boomers exit the workforce than was happening annually pre-lockdown. Not long before the pandemic, a common refrain was that boomers were staying in the workforce too long, longer than their parents did, and thus robbing millennials of jobs and advancement opportunities.
Last year also saw a record number of employees of all ages simply quit. Compared with boomers, nearly twice as many millennials—now the largest share of the workforce—say that they plan to quit or change jobs. While Gen Z remains a small share of the workforce, over half of that growing cohort claims that they’re thinking about quitting, too. Add to this the now record-low labor-force participation. That basic measure, in decline for decades, had stabilized for the half-dozen years prior to the lockdowns.
These trends have produced an historic labor shortage. This year began with some 11 million job openings but only about 7 million unemployed citizens looking for work. That’s a complete reversal of the state of affairs of the first 15 years of the twenty-first century, when those seeking work outnumbered available jobs. And even though significant, the migration of citizens into self-employment—especially the Cloud-enabled “gig” workforce—doesn’t come close to explaining the labor shortage gap.
No corner of the economy has escaped this new reality. Staff shortages in restaurants are nearly universal. The 2021 data show more than 1.5 million unfilled job openings in America’s “accommodation and food services” sector. Similarly, in the supply chains, we see a shortfall of 80,000 truck drivers, a gap forecast to double by 2030, with over one-fourth of today’s truckers over the age of 55. Meantime, warehouse operators are in hot competition for the same people who might drive trucks, or work in retail, or in manufacturing, including in the cardboard-box industry, which itself faces an epic labor shortage. So, too, the oil industry, which fuels it all, sees demand already above pre-lockdown highs while facing a deepening talent shortage.
The labor gap is hardly isolated to “blue collars.” Semiconductor manufacturers and cloud infrastructure providers are in a war for talent, with labor shortages amplified now as both industries plan unprecedented expansions. The health-care sector, with a 50-year-old average age for doctors and nurses (demographically similar to trucking), is seeing exits accelerated by what can only be called Covid burnout. One-fifth of hospitals report critical staff shortages, with nearly half of nurses saying that they’re considering quitting soon.
Some of these trends—eagerness to quit, say—may moderate. The same nursing survey found that 60 percent might stay for higher pay. As Fortune headlined it; “The Great Raise is the solution to the Great Resignation.” The advice-giving industry is replete with formulae for retaining employees, both boomers and millennials. In addition to money, employers are exhorted to make workplaces more fun and attractive (hybrid, of course), and to adopt management behaviors that are, in a word, nicer.
But many of the trends, especially demographic, are intractable. A birth dearth, combined with a worrisome share of citizens that just don’t seem to want to participate in the workforce, will yield a future with a persistent shortage of increasingly expensive labor for all businesses. Employers will be forced to do less, constraining growth, and even then, at higher costs; or they will need to find technologies that amplify what the available employees can do.
Enter the age of the robots, especially service bots.
Thus far, most of the twenty-first century has been dominated by increasingly hyperbolic claims that robots and artificial intelligence (AI) will soon take over so many tasks that millions of citizens will be put out of work and have no choice but to take a proposed “universal basic income,” never mind old-fashioned short-term unemployment checks. Paying for that would come from taxing the businesses profiting from automation—and even taxing robots.
This narrative has many flaws. The biggest is that it has reality backward. Robots and AI—the latter a virtual robot in the Cloud—are precisely what’s needed to fill an expanding labor deficit, and for boosting wages. Fortunately, recent trends show that useful robots are, finally, becoming practical.
It’s worth restating the purpose of automation: it’s to get the same or more output using fewer inputs of labor, materials, and money. That, by definition, is productivity. And that, history has shown, is the key to expanding society’s overall wealth, including higher wages. An employee paid, say, 50 percent more but who is 60 percent more productive yields both lower cost outputs for consumers and higher profits for businesses, a proverbial win-win. As economist Joel Mokyr wrote in his seminal book, The Lever of Riches, technology is the one thing that can create a “free lunch.” Or as Musk put it, “capital equipment is distilled labor.”
The chief problem with automation—and robots, especially—is that for the century since Henry Ford’s first mass production line, humans do a better job for most tasks, most of the time. Put inversely: robots haven’t been good enough, either in performance or costs, to take on most tasks, especially in collaboration with people. Workplace psychologists have made a career of studying and counseling on employee collaboration. But the technologists have had a harder time inventing robots that play well with others, whether it’s other people or other machines. These are the enduring challenges of making it easy to install and operate any machine (the “user interface”), as well as the need for, in engineering jargon, “interoperability.”
But now the long over-promised arrival of useful robots is being fulfilled. It took decades of progress in key enabling technologies. The evidence of that progress is visible in trends that predate the Great Resignation.
Consider where it all began, with factory robots. The ratio of those bots per employee—the “boots on the ground” metric—has doubled in just five years. Overall global spending on all robotics and artificial intelligence (combined) rose 400 percent in the first two decades of this century. Even before accounting for the new labor trends, forecasts now expect spending to accelerate and double in just five years. Far more portentous: for the first time, last year saw services, not industrial applications, make up over half of all global robot purchases.
Nowhere is the service bot more needed and more obviously ready for prime time than in the warehouses at the heart of supply chains.
It escaped no one’s attention that the lockdowns lit up e-commerce, with utilization rising in one year as much as it had over the previous half-dozen years. That was made possible not only by shopping in the Cloud, society’s newest infrastructure, but also by a torrid expansion of warehouses. The warehouse, an infrastructure of commerce as old as civilization, is now the locus of the most rapid influx of robots anywhere. In the coming decade, more robots are expected to be “hired” by warehouse operators than in all other applications combined. While the $16 billion spent on warehouse automation last year is 60 percent higher than five years ago, it’s now expected to increase by more than double over the next five years.
Professional cleaning and health care are the other two fast-growing markets for service bots. Improved performance, adaptability, interoperability, and lower costs are allowing robots to invade every service sector—from security and safety to construction and restaurants. At least one analysis sees the overall market for service bots by 2030 reaching from $90 billion to $200 billion. Adding in industrial bots takes the totals to $160 to $260 billion.
Such forecasts are based on trends already underway. For example, Tyson Foods recently increased wages and benefits and is also investing $1 billion in automation not only to fill open positions but also to upskill employees. Restaurateurs are starting to install robots that can cook French fries, flip burgers, or pour drinks (Pitchbook has flagged four dozen companies making these beverage bots), freeing up staff time to attend to patrons.
Similarly, robots that can unload boxes from trucks will shortly free up people to either do other work in the warehouse or drive the trucks. (Robotified trucks are going to take far longer to mature than popular conception suggests.) Boston Dynamics, famous for its YouTube videos of dancing and backflipping robots, recently introduced a box-handling bot that can finally match the 800 box-per-hour rate at which humans unload a truck, and needs to take a break only every 16 hours (to recharge). DHL Supply Chain placed the first order for a small fleet of them. Robots that can pick fruit are also finally available, itself a long-standing challenge because of the delicacy of that task.
And the same technology gains that make bots useful in handling boxes or delicate fruits are also expanding the robotification of the scientific research laboratory—not just filling labor gaps but amplifying what technicians and scientists can do.
That all of this has happened only recently is not the consequence of some specific invention or epiphany—nor, for that matter, any government program or subsidy. Engineers have been trying to build general-purpose, free-range robots for a long time. The first (practical) design for a modern, untethered, and ambulatory robot dates back more than three decades, to a 1989 paper published by MIT roboticist Rodney Brooks. Then, in a 2008 Scientific American article, Bill Gates got out “over his skis” predicting that robots were on the verge of appearing in everyone’s home, analogizing this coming development with the PC. But software, Gates’s focus, was only one of the challenges in making robots useful. Revolutions were also needed in sensors and power.
Sensors for vision and location took longer to advance, but in due course they, too, have followed a pattern similar to the often-noted Moore’s Law for computer chips. Roboticists now have available an array of options in the form of tiny, powerful cameras, chip-scale radar, complementary laser-based radar (lidar), as well as microscopic, silicon-fabricated position sensors. Sensing motion, direction, and velocity is intuitive for humans but hard for machines to emulate. The inertial measurement unit, or IMU, that can detect changes in movement, has been used by the military, in particular, for a long time. But only in the past two decades has the IMU shrunk from coffee-cup scale to chip-scale, and only in the past decade has it gained both sufficient precision and affordability.
Practical, tether-less robots also needed a revolution in power, both to store onboard energy and to enable actuators that can effect movements and manipulations with precision. The first (game-changing) commercial lithium battery wasn’t invented until two years after Brooks mapped out a useful robot design, and it took another decade or two for the requisite maturity. Similarly, actuators—in effect, robots’ muscles—have followed another, independent trajectory of (fortuitous) advances in size and power. Over the past several decades, superior designs and new materials—especially the 1984 invention of rare-earth neodymium super-magnets—have engendered a roughly fifty-fold gain in the power-to-weight ratio for tiny electric motors.
Far more software horsepower was also needed to put the data tsunami from the new classes of sensors and actuators to work. (About that, Bill Gates was right.) It has taken decades for computers and networks to become fast enough, and cheap enough, to manage the scale of data associated with modern sensors and actuators. Speed and precision are critical for autonomous navigation in unconstrained environments—in other words, instead of being bolted to the floor in cages in factories or following predetermined pathways.
The revolutions in “machine learning” and AI, bringing blazing speed to analytics for physical robots, are, of course, the same tools increasingly put to work in the automation of “knowledge work.” The new domain of so-called robotic process automation (RPA) takes on a wide range of business processes, not physical tasks. RPA automates the many varied types of business processes from, say, on-boarding employees to monitoring safety, or making real-time decisions about goods in the supply chain. Bringing RPA to knowledge work is as significant as going from pen-and-paper to spreadsheets.
But the technical community has adopted a confusing set of terms for this class of software, including intelligent process automation, business process management, cognitive automation, or just artificial intelligence. The more recent overarching term “intelligent automation,” or IA, is both more accurate and less frightening, and less misleading than the term AI. IA may sound like a tautology, but it describes an evolution from simplistic automation typical before the advent of the Cloud to systemic automation integrated with real-time networks, and situational awareness through sensors.
As with box-lifting robots in warehouses, the robotification of business processes—the knowledge work—is not notional but well underway. At least one survey found that an overwhelming majority of businesses already use or plan to use various forms of IA for administrative tasks—the tasks that most employees classify as the drudgery part of their work.
IA is already a $20 billion industry that’s projected to double in just a few years. Despite what some pundits imagine, employees embrace software that is useful and employs an intuitive “user interface,” and that is easy to integrate with other tasks, people, and other types of software (i.e., real-world “interoperability”). Think of the capabilities as akin to Siri or Alexa on steroids, combined with the frictionless functionality of consumer apps like Uber or Airbnb.
Today’s automation forecasts rarely account for the impact about to come from perfecting augmented reality (AR) and virtual reality (VR) . While still predominantly used in games, both technologies are finally maturing in ways meaningful for business applications. Again, we don’t have to imagine whether better tools are possible since AR glasses that are useful (rather than geek-like) are now emerging or planned, including fashionable ones promised from, for example, Ray Ban, in collaboration with Facebook (now Meta), or soon from Apple, Google, and dozens of startups.
Few areas will be more productively affected by AR/VR—the so-called metaverse—as will health care. Today’s still-clunky telehealth has, nonetheless, been productive. As many discovered during the lockdowns, telehealth—the utilization of which has jumped over thirtyfold—is clearly labor-saving; it just needs to be technologically improved. That’s happening now, and fast.
Still, count on the “Zoom class” of writers and pundits to continue to claim that robots and IA will mean the end of work. And, for reasons of inscrutable human psychology, it’s the anthropomorphic robot—ones that can pick up boxes, pour drinks, clean rooms, deliver medicines, or help a nurse lift a patient out of bed or off the floor—that most animates the punditocracy to predict a jobs apocalypse. That won’t happen because it’s never happened. Automation—enhancing human productivity and amplifying human creativity—has for more than a century led to both a transformation in and an expansion of available work. Robots are merely, even if profoundly, the next step.
The birth dearth will take time to reverse. But coming fast now are hundreds of new kinds of robots from both startups and established companies. Doubtless, both businesses and citizens will become more animated if, as he did with the electric car, Elon Musk makes good on his promise to build a Tesla robot. As he said recently, it would be his company’s “most important product.” And he would be right.
The wheels of robot progress are in motion, and none too soon.
City Journal is a publication of the Manhattan Institute for Policy Research (MI), a leading free-market think tank. Are you interested in supporting the magazine? As a 501(c)(3) nonprofit, donations in support of MI and City Journal are fully tax-deductible as provided by law (EIN #13-2912529).