A REVOLUTION is occurring at the intersection of technology and labor. Change will be uneven across sectors and across the globe. But it will come. Technological advances in artificial intelligence (AI), autonomy and robotics will fuel these changes, aided by the synergistic effects of the Internet of Things (IOT), three-dimensional (3D) printing and big data sciences.

The development of these technologies brings the promise of higher productivity (and, with it, economic growth), increased efficiencies, safety and convenience. However, it also raises difficult questions about the broader impact of automation on jobs, skills, wages and the nature of work itself.

The effect on labor will undoubtedly be disconcerting for many as jobs will be eliminated or greatly evolve. Some “traditional” jobs will become obsolete or numbers of workers in them reduced as technology further encroaches into the workplace. In addition, workplace productivity will increase, while the human labor demands will be dramatically decreased.

Some futurists, such as Homi Kharas at the Brookings Institution, predict that a child born in 2020 might never have a job. Others talk of different types of jobs—nontraditional by today’s standards—where workers will work on a contingent basis. Examples of early adapters in this type of work are Uber and TaskRabbit, where workers set their own schedules and decide how many hours to work. A World Economic Forum report, meanwhile, estimates that “65% of children entering primary school today will ultimately end up working in completely new job types that don’t yet exist.”

While some question the scale and pace of the changes, near-unanimous agreement exists that whatever portion of the workforce remains will require new skills. Thinking will likely replace doing for much of the workforce. Positions involving routine or repetitive skills will likely be the first to be eliminated. A premium will likely be placed on those workers with specialized skills, particularly in the science, technology, engineering and mathematical (STEM) disciplines.

If history is any indication, fewer people will be involved in traditional work: agriculture, industry and services. So what will all these people be doing? And what, then, is the future of work?

THE CHANGES underway in labor markets are breathtaking, but not unprecedented. Throughout recorded history, an inverse relationship has existed between technology and labor—as the use of technology has increased, labor demands have decreased, productivity has increased and restructuring of the labor markets has resulted.

Every advancement in human history has sprung from the desires to acquire safety and security, to learn more about the world and to improve human welfare. As humans continued to advance and build societies, functioning economies took root, and with them the idea of work. Throughout, even when disruptive technologies have been introduced, societies have found ways to adapt.

During the Industrial Revolution, from 1750 to 1900, technological developments such as the steam engine, telephone and railroads began to transform mostly agrarian societies towards greater industrialization. From 1900 to 1950, electricity and mass production led to greater productivity and more rapid technological advances. The development of computers and information technology (IT) since the middle of the twentieth century ushered in the Information Age. In the seventy-year period from 1900 to 1969, humankind went from horse travel to landing a man on the moon. With each of these transitions, an accompanying increased productivity in agriculture, industry and the sharing of information resulted.

Great benefits have accrued from these advances, but sizable shifts in labor markets have also occurred to challenge traditional thinking about the workforce. In 1840, approximately 70 percent of the workforce was engaged in agriculture, with the other 30 percent working in industry or services. By 2012, the percent of the workforce in agriculture stood at 1.4 percent. Industry—including mining, construction and manufacturing—rose steadily from 1840 to 1955, peaking at approximately 38 percent of the workforce. Since, it has been in decline, and now accounts for less than 20 percent of the U.S. labor market.

The service sector has continued to grow steadily from 1840 to 2010 from approximately 20 percent to almost 80 percent of the U.S. labor market. However, it is unlikely that services can continue to experience this pace of growth as a percentage of the U.S. economy, as that would imply continued complementary reductions in agriculture and industry.

Factors influencing the decline in industrial labor requirements vary by sector, but some common themes have emerged. Efficiencies have been gained through the development of automation and robotics. Efficiencies have also been gained through optimization efforts, such as lean manufacturing techniques, which have further reduced demand for workers.

In manufacturing in particular, economic restructuring is making the need for large labor pools of highly specialized craftsmen obsolete. Instead, the combination of technology and specialized techniques has allowed smaller boutique manufacturing operations to flourish. These operations offer increases in speed and precision while lowering labor costs.

Offshoring of manufacturing—including related support services, such as information technology, finance and accounting, procurement, and marketing, to name a few—has also taken a toll. Cheap labor and the U.S. tax code incentivized moving production, jobs and profits overseas by favoring overseas investment with a lower effective tax rate.

While it is unlikely that the industrial sector will experience the same percentage of reduction as agriculture did, a large rebound in the manufacturing labor market also seems unlikely. Under certain conditions, some industry could return to the U.S. economy, but certainly not the large-scale, labor-intensive manufacturing plants of the last century.

Although labor trends have seen a reduction in the percentage of the population in agriculture and industry, production and output are at their highest levels. For example, according to data from Pew Research, employment in manufacturing was 17.5 million in 1987 and 12.4 in June 2017, approximately a 30 percent reduction. During that same period, the seasonally adjusted real output index grew from 69.79 to 129.26, an increase of 85 percent.

The service sector has been on a steady climb since the 1800s, but can this growth continue? History suggests that would be very unlikely. Just as in agriculture and industry, efficiencies through technology are likely to be identified, labor markets altered and the service workforce consequently shrunk. The leading edge of these changes can be seen with the growth of companies such as Amazon and Uber that are disrupting labor markets.

A RESTRUCTURING of the global economy fueled by changes in technological capabilities is underway. These changes will result in a complementary restructuring of entire sectors of the global economy. They will be felt most dramatically in the United States and other advanced nations in the coming decades. While the benefits of these shifts may not be seen in developing nations for some time, the negative effects of this economic restructuring will be felt almost immediately.

Technological advances in artificial intelligence, autonomy and robotics, and enabling technologies such as IOT, 3D printing and big-data science, will alter fundamental relationships—and ultimately the cost-benefit calculus—between technology and labor costs, global transportation network costs, and consumer costs for goods and services. Technology will be used to offset requirements for human labor. Global transportation networks are likely to contract, as advanced manufacturing techniques and 3D printing will mean goods can be more efficiently produced in local or even neighborhood manufacturing centers. Lower labor costs could translate to lower consumer prices, although lower cost to market may imply higher profits rather than savings for the individual consumer.

In examining the future of work, dire predictions have become commonplace. Books such as Humans Need Not Apply and Rise of the Robots chronicle a jobless future in which increasingly traditional human tasks are outsourced to technology. Given these predictions, it seems prudent to consider the extent to which existing jobs can be automated.

Consider automobiles: autonomous-vehicle technology could dramatically alter our interactions with cars. One account in the San Francisco Chronicle optimistically predicts autonomous vehicles will “prevent accidents, ease traffic and bring newfound mobility to the elderly and the disabled.” It could also save 1.2 million lives annually—forty thousand in the United States alone—from losses due to vehicle accidents.

But autonomous vehicles could prove highly disruptive as well, altering entire industries. Long-haul trucking is a prime example, ripe for an infusion of autonomous technology and a move away from human drivers. Imagine the benefits in shipment times and costs savings for autonomous fleets of trucks that could drive twenty-four hours a day, not requiring rest for the drivers. An important secondary benefit would be increased highway safety.

The trucking industry includes approximately 3.5 million professional drivers and 5.2 million more in the industry in nondriving positions, or a total of 8.7 million trucking-industry jobs. But that’s only the tip of the iceberg: highway restaurants and other businesses also exist to support the trucking industry. While an exact number of people supporting the long-haul trucking industry would be highly speculative, it is safely in the millions. If, for argument’s sake, we estimated the total in or supporting the trucking industry to be fifteen million people, that would amount to almost 5 percent of the U.S. population or 10 percent of the 153-million-strong U.S. labor force, according to the Bureau of Labor Statistics. While the local trucking industry is likely safe for now, the repetitive and dangerous long-haul part of the business is a potentially lucrative target for a labor-technology conversion.

The trucking industry is not the only potential victim. Driverless vehicles could change car ownership trends, making full ownership an obsolete concept. Alternatives could include fractional ownership, own-and-share, mobility services or paying per ride. Imagine the effect on car dealerships, auto mechanics and even parking attendants if people were to forgo the traditional models of car ownership and commuting. While conversion to 100 percent autonomous vehicles might not be possible, even in the long term, it is safe to predict that a large percentage of cars will eventually be driverless.

Manufacturing continues to be a target for greater use of technology, particularly automation, robotics and AI. Technology comes with the temptations of lowering labor costs, increasing throughput and improving quality. While robots may not supplant all human positions—at least not yet—their continued proliferation throughout manufacturing is assured. America has lost approximately seven million manufacturing positions since its peak in 1979, largely due to technology, yet output has continued to increase.

Simply put, the numbers are staggering. General Motors has lost one-third of the six hundred thousand workers it had in the 1970s, and is outproducing its previous totals. The steel and metals industry has lost over 265,000 jobs, or 42 percent, yet production has increased by 38 percent. One estimate is that by 2020, the United States will be the “most competitive country in manufacturing.”

The same is true for mining, where the labor force has been reduced from over 250,000 to 58,000, due to new ways of mining that require less manpower, thus effectively outsourcing positions to technology. Despite these labor reductions, output continues to grow.

The so-called Internet of Things could affect a wide range of industries. From monitoring industrial controls systems for utilities to health-care delivery, opportunities for reducing labor are evident. Anywhere workers are performing checks, monitoring statuses or adjusting levels has the potential for automation, and ultimately AI. Some might argue that certain tasks require human skills, but, as events like the Chernobyl and Fukushima nuclear accidents demonstrate, human mistakes can cause or contribute to devastating outcomes. Both would have benefited from AI-equipped nonhuman intervention rather than relying on human judgment.

PERHAPS NO sector has greater potential for dramatic change than retail businesses and supply chains. Today, if a customer wants to buy an item—say, a plastic bowl—she can make a trip to the store and purchase the bowl. However, that bowl has traveled a long way to reach store shelves, where the customer picks out the item based on its size, color and shape. It was likely made in a manufacturing facility overseas, shipped by ground conveyance to a port where it was loaded into a crate, placed on a ship for transit across the ocean and ultimately delivered in a reverse process to the store. Of course, the nature of retail stores is to provide choices for customers, which implies that businesses should keep their shelves filled, shoulder the costs of holding inventory and sell products at a low enough price that inventory will move quickly.

Now let’s consider what 3D printing will do to the retail businesses and supply chains of the future. The potential customer now goes online and orders a bowl, able to alter the dimensions to suit her individual preferences. A local 3D manufacturing facility (which has not yet been built) receives the order, prints the bowl, packages it for shipment and places it on a drone that has been configured to deliver packages. The entire process could be completed in hours vice the several months’ journey the first bowl took to reach the retail store. Theoretically costs could be reduced, as the product takes fewer steps, fewer people are involved in “delivery” and there are no inventory holding costs.

This is great for the customer with the new bowl fitting her exact specifications, but what about the workers in the long supply chain that stretched from the overseas factory to the retail store? Having fewer people involved in the process means fewer workers overall. In fact, the early signs of this supply-chain revolution can already be seen in the closure of stores whose business models rely on long supply chains and large inventories.

Of course, as the manufacturing sector evolves, not all positions will be eliminated. Some undoubtedly will result in new positions, such as in 3D manufacturing plants and aerial-delivery service centers. Still, future labor markets are likely to experience volatility while reaching a new normal, and as labor demands are reduced or new workforce skills acquired.

Returning to the question of how to determine what positions are likely to be automated, the answer varies around the globe and from sector to sector. However, trends are uniformly leaning towards a greater reliance on technology to offset human labor demands across virtually all sectors. Today, estimates from McKinsey suggest that some 60 percent of all jobs include at least 30 percent tasks that automation could replace, based on currently demonstrated technologies. A separate study from Oxford University found that of 702 occupations considered, 47 percent of U.S. workers had jobs at a high risk of potential automation.

As technologies mature and machines gain higher levels of performance that meet or exceed human capabilities, displacement of today’s workforce will continue. The McKinsey report suggests that automation technologies could affect 50 percent of the world economy, or 1.2 billion employees and $14.6 trillion in wages. When these people are displaced, what will they do to earn a living and provide for their families?

And what determines whether a job is at risk for automation? Several factors serve as indicators for identifying jobs at risk. First, a job’s technical requirements provide some important indicators. Is it routine or repetitive? Does it require thinking and judgment, or routine activity with little need for creativity? Second, do technical impediments exist for developing the hardware and software required to automate a position? The response likely varies over time as additional technological sophistication is achieved. Perhaps today, with the current level of AI capacity, the answer is no, but what about in the future? Third, does the labor market provide a ready supply of labor at an affordable cost? Does the cost of automation far exceed the cost of maintaining human labor in the position? And, fourth, are there benefits to be gained by automating the tasks. For example, can a dangerous task be automated to protect workers from injuries? Will higher levels of output and quality be achieved? Can an automated system surpass human abilities?

Many are quick to point out that technology and associated automation is not an all-or-nothing proposition. Some functions that have been automated have resulted in the growth of other jobs. For example, per McKinsey, while the internet is credited with eliminating five hundred thousand jobs in France over a fifteen-year period, it simultaneously created over 1.2 million new ones. Likewise, the emergence of the internet economy generated a greater need for statisticians and data scientists.

Still, on balance, trends towards greater use of technology in place of labor are continuing, and will alter future labor markets. As the Economist summarized the Oxford study’s conclusions, “the workforce bifurcates into two groups doing non-routine work: highly paid, skilled workers (such as architects and senior managers) on the one hand and low-paid, unskilled workers (such as cleaners and burger-flippers) on the other.”

THE FUTURE of work has become a topic of serious debate. Some foresee the society of the future as a jobless one in which all work is done by robots. Others take a more measured view, allowing that labor markets will change incrementally, but affirming that societies will be able to adapt and reach a new equilibrium. Those in the incremental camp highlight the need for more STEM education, retraining of people displaced by technology, and safety nets and transition support until they can reenter the job market.

These approaches sound well reasoned and certainly prudent in the near term. However, what if the direr assessment that technology will displace 1.2 billion jobs and $14.6 trillion in wages came to pass? Such measures would likely not be sufficient.

Strategies would then need to be larger and require more resources. They could include the following:

Creation of a robot tax. Bill Gates advocates a robot tax, “as a way to at least temporarily slow the spread of automation and to fund other types of employment.” He calls for governments to use revenues to develop safety nets for the displaced workforce, and perhaps even other social programs. The proposal has been hotly debated, with many saying that the tax would adversely affect innovation and productivity. The European Union has proposed a similar initiative.

Changing the structure of work. Today, work is primarily focused on output. As such, in agriculture, industry and services, workers produce a tangible output that is assessed to have a certain market value for which they are compensated. Very few people work in think tanks, research laboratories and other idea factories. However, if technology has displaced workers in routine and repetitive tasks, perhaps that brainpower could be put to use solving complex, seemingly intractable problems that are or will be facing society. One could see this as an opportunity to create a new age of enlightenment, focused on hard problems and generating ideas.

Renewed focus on research and development. R&D has been an engine of national growth throughout America’s history. However, the share of R&D taking place in the United States—both executed and sponsored by government—has been steadily declining. The displacement of the traditional workforce could be an opportunity to reinvest in future growth. Despite progress across virtually all scientific disciplines, much work remains to be done. Consider biology, where we still understand only a tiny percentage of the function of the human genome’s 3.3 billion base pairs, and emerging infectious diseases remain a daily threat in much of the world. Or advanced materials, where self-assembly offers solutions in areas as diverse as infrastructure and personalized medicine. Progress in R&D could even lead to greater understanding of natural disasters, risk mitigation and the future of resilient societies.

Greater corporate responsibility. Corporations have a vital role in the future of technology and labor. Rethinking how companies measure their success could be a step in the right direction. A relatively new corporate structure could provide such a path: a benefit corporation officially bases its success on measures beyond traditional ones like quarterly corporate earnings. A benefit corporation, as one introduction to the concept explains, “legally protects an entrepreneur’s social goals by mandating considerations other than just profit.” If a benefit corporation begins to prioritize societal benefit as much or more than quarterly earnings, labor enrichment could come to be valued as much as return on investment.

Developing a national strategy for the future of work. The public and private sectors must work together to ensure a viable future workforce. They must address the full range of economic, security, social, and health and welfare outcomes associated with large pools of untapped labor. Actions taken now could help mitigate, to the extent possible, the coming displacement of jobs. A laissez-faire approach that allows market forces to drive future labor requirements will likely result in a less prepared future workforce, economic instability and perhaps even societal unrest.

THE FUTURE is not preordained to be a workerless society in which output is generated solely by AI- enabled robots “led” by C-suite managers. Disruptive technologies and innovations have displaced workers in the past, and society has weathered the storm.

However, this time feels somewhat different. Technological advances in AI, autonomy and robotics—being aided by the synergistic effects of the IOT, 3D printing and big-data science—are fueling these changes. In other words, technology is coming to substitute for doing and thinking, and increasing the number of jobs that might be made obsolete.

Opportunities exist today to begin a serious national dialogue about the future of work. Our ability to seize these opportunities will be key to navigating the future that lies at the intersection of technology and labor.

Daniel M. Gerstein works at the nonprofit, nonpartisan RAND Corporation and is an adjunct professor at American University. He was the undersecretary (acting) and deputy undersecretary in the Science and Technology Directorate of the Department of Homeland Security from 2011–14.