The employment marketplace faces a persisting churn of automation, AI, and robotics displacing human workers. Some pundits disagree on the severity of the disruption Marc Andreesen says current alarm over robots in the workforce is overblown. Bill Gates and Mark Zukerberg agree that the displacement will occur, but they offer a silver lining to new college graduates who might refocus their career into helping the less fortunate.

At a Harvard commencement, Zukerberg said, “Our generation will have to deal with tens of millions of jobs replaced by automation, like self-driving cars and trucks. Technology and automation are eliminating many jobs.” He noted that the change could undermine our sense of purpose and value, but that we could intentionally repair that loss through propose-driven employment. Bill Gates offered similar observations; “robots should free up labor and give graduates an opportunity to focus on jobs that only let us do a better job of reaching out to the elderly, having smaller class sizes, and helping kids with special needs.”

Estimates of the quantity and timing for robot-displacement of jobs support Gates’ and Zukerbergs’ expectations of job losses. Forrester Research expects robots to create 15 million new jobs in the U.S. over the next 10 years as a direct result of automation and artificial intelligence. Robotics will also kill 25 million jobs over the same period. That’s a net loss of 7% of the workforce.

The displacement in employment will not be evenly spread. On the losing end, MIT and Boston University estimatedsix workers will lose their positions for every robot added.” Workers performing “routine manual labor or assembly-line production were expected to be hardest hit, alongside people without a college education.” We have seen this in Foxconn, one of China’s electronics factories, where robots will replace most of the factory workforce. Robots can run 24/7 without losing production time for specialized concessions that humans expect (paid maternity leave, sick days and vacation days).

To understand the scale of robot replacement of employees, we will need to determine the number of workers displaced by robots, year by year, and the number of workers who have or who acquire the skills to fit into other jobs, year by year. The displacement and new job processes result in a churn within the employment market, out of routine jobs and into jobs that need higher skills. While this renewal in the labor market is underway, some workers will be retiring, some will reenter the market as employees or self-employed, and some will be sidelined for long periods. Robots will not replace 100% of workers, but some workers will be displaced permanently, and those with new jobs may experience a change in income.

The cycle of renewal depends on strong basic education, relevant majors in university, and effective retraining for both employed and displaced workers. To qualify for new jobs, workers will need good communication, supervisory and technical skills. To offer new jobs, the economy will encourage the creation of new businesses or expanded operations that compete well against trading partners. The gains from robot “labor” will flow to the robot owners, not to the aggregate workforce. This may require some rethinking of taxation.

If there is a persisting shortage of skilled workers in the workforce, some highly educated candidates may be drawn back from retirement. The $400 trillion deficit in pension funding and their extended lifetimes will motivate some to return to the job market. Quality of K-12 education and worker retraining are important. It will be tempting to adjust taxation to accommodate the new employment mix, but the era of human worker displacement would be a foolish time to consider expanding entitlements.