Many people are concerned about artificial intelligence eliminating jobs, but they may be overlooking its likelier and more subtle impact: job churn.

As companies increasingly adopt machine learning, they'll need fewer full-time employees for certain tasks, resulting in more part-time jobs or temporary work—all without health and retirement benefits.

“In an economy where benefits are tied to full-time employment, any increase in job churn would create instabilities and insecurities in people’s ability to maintain their income and their health and retirement benefits,” said Brookings Institution director of governance studies Darrell West last week during a U.S. House Budget Committee hearing about A.I. and the workforce.

The situation would be akin to aspects of the so-called gig economy, in which workers juggle multiple jobs of driving people around for Uber and then delivering dinners for DoorDash.

It’s unclear how the U.S. would deal with laid off or under-employed workers who are unable to find new jobs because of their lack of technical training. Countries like South Korea and Germany that have experienced technological upheaval were able to create new high-tech jobs when some older ones disappeared, explained Massachusetts Institute of Technology economics professor Daron Acemoglu. That hasn't been the case in the U.S., he said.

He drew a parallel between the impact of A.I. on jobs and the U.S. auto industry’s shift to more automated manufacturing. Production jobs disappeared in cities like Flint, M.I, and nothing replaced them.

While he didn't mention it, perhaps countries like South Korea and Germany have fared better than the U.S. because they have instituted large-scale job training programs.

Speakers during the hearing floated several ideas about helping workers in the machine learning era. West said the U.S. should improve Internet access for people living outside of metropolitan areas for better access to online job training and education services as well as to help them work remotely, as more companies shift their operations in light of COVID-19.

Participants also expressed the need for schools to focus more on areas like computer science and math, described as a key to workers succeeding in A.I. West noted the crucial role community colleges play in providing workers with skills more cheaply than four-year universities. He also advocated for the creation of so-called life-long learning accounts for workers, akin to a retirement fund but focused on education so people can receive training throughout their careers.

Indeed, there may be no way to stop the rise of machine learning, but there’s still time to prepare.