Concerns about technology-driven mass unemployment are not new. The earliest fears date back to the time of the Luddite movement in England when laborers attacked stocking frames — an early knitting machine — because of the displacement of traditional hand-knitters.

Then in 1930, economist John Maynard Keynes predicted a new disease: Technological unemployment as discoveries of labor-saving technologies outpaced the discoveries of new uses of human labor. Then again in the 1960s, President Lyndon Johnson commissioned the Blue Ribbon National Commission on Technology, Automation and Economic Progress to investigate whether technology was on the verge of putting too many people out of work.

They concluded that technology influenced the places where work was performed, the kinds of industries organized around the work and the people affected by unemployment; but the overall demand for goods and services determined the amount of work to be done. In short, they concluded that technology eliminates jobs, not work. Today’s techno-pessimists would agree with the commission and argue that history is on their side. Despite technological disruptions, humans have always found more work to do.

MIT’s David Autor, a leading observer of automation’s impact on the labor force, points to several reasons for labor’s persistence.

Technology complements labor as much as it substitutes for it

Increasingly, today’s occupations bundle a number of tasks — some non-routine, some routine — requiring a combination of creativity, intuition and judgment, mixed with memorization and repetition. Automation can substitute for the routine tasks and shift the emphasis of the occupation to non-routine tasks.

Bank tellers and the development of automated teller machines (ATMs) provide a great example. ATMs were first introduced in the U.S. in the 1970s and quadrupled in number between 1995 and 2010 from 100,000 to 400,000. Over a similar time, between 1980 and 2010, bank teller jobs also increased from 500,000 to 550,000. Rather than destroy the occupation, technology changed it. The mix of tasks performed by bank tellers shifted from routine cash handling and account reconciliation to an increased focus on client management, sales and problem solving. Banks deployed tellers to strengthen relationships with their clients.

Humans know more than they can tell

Automation will be limited by Polanyi’s paradox, named after the economist/philosopher Michael Polanyi, which asserts that many tasks draw on tacit, intuitive knowledge that is difficult to write down or codify. Human drivers use judgment and common sense when they encounter surprises on a roadway. For example, different reactions occur for a downed power line than for a tree branch. The qualities of leadership are hard to teach or fully explain, but we know it when we see it.

We can identify great teachers but would have a hard time documenting all the specific qualities and actions it takes to create one. Architects, designers and artists rely on an aesthetic sense to build and create work that is appealing to others. Body language can be as important as the spoken word in communication, but is hard to train. Emotional intelligence is critical in all work settings, but is difficult to document. Computerization and robotics have begun to break into tasks previously believed to be off limits, and technologists believe that machines will eventually break into tacit knowledge through massive amounts of testing, trial and error. How far they get, and by when, will determine the degree of labor market disruption.

A future of mass automation and greatly improved AI requires investment. The technology doesn’t build itself — at least not yet. Here, broad market indicators do not suggest an imminent wave of automation. Business investment in information technology and software — expressed as a share of the economy — fell in the aftermath of the 2001-era tech bust. Despite all the promises of a technology-driven economic revolution, labor productivity has been on a long, steady decline since the end of World War II. A surge in business investment and a corresponding rise in productivity could signal labor disruption, but the signs aren’t there, yet.

Widespread dissemination of some technologies will also require policy innovation, political acceptance and public-sector investments. The deployment of autonomous vehicles or drone deliveries in complex urban environments poses a wide array of infrastructure and land-use questions that cities are just beginning to ask. What is technically feasible will collide with what is practical to implement. As in the past, political barriers and bureaucratic processes will slow the pace of change.