The implications of machine learning are wide-reaching — and that includes the regions of the U.S. that’ll be affected job losses due to AI. In a surprising turn, a new report suggests that the areas that’ll be hardest hit by automation in the future aren’t the ones full of manufacturing jobs, according to this piece on The Atlantic. It’ll be those dominated by jobs in food preparation, office and administrative support and sales.

That leaves Las Vegas and the Riverside-San Bernardino area at the top of the list for job losses. The report, from the Institute for Spatial Economic Analysis at the University of Redlands, found that 65 percent of jobs in Las Vegas and 63 percent of jobs in Riverside will be automatable by 2025. “We felt it was really stunning, since we are underestimating the probability of automation,” Johannes Moenius, director of the Institute, told Atlantic writer Alana Semuels. El Paso, Texas, Orlando, Florida, Louisville, Kentucky and Greensboro, North Carolina are also vulnerable to high rates of job loss due to automation.

Semuels explains that though much has been made of the threat that automation poses to manufacturing jobs, many of those positions have already been automated (particularly in the Midwest), so “those regions are not at the top of the list of the places that currently stand to lose the highest share of jobs.” Instead, automation is now “coming for lower-income jobs,” Semuels writes. In the not-too-distant future though, that’ll likely expand even more as machines learn to do less-routine, more nuanced tasks.

The areas that are poised to see the least job loss due to automation? Silicon Valley, Boston and the Research Triangle in North Carolina, where, as Semuels writes, “a high share of the jobs require more creative and social intelligence, and are thus more difficult to automate.”

That brings us to a point we’ve covered before: The jobs that most require a human touch, whether it’s creativity, social know-how or emotional intelligence, are the least likely to be taken over by machines. That tracks with what the researchers concluded, too: According to their findings, “the livelihoods of mental-health and substance-abuse social workers, oral surgeons, choreographers and physicians were more protected,” Semuels writes.

As AI and automation spread and machine-learning becomes capable of more sophisticated tasks, emphasizing what your humanity adds to your job performance may be the smartest move you can make.

Read more on The Atlantic.

Originally published at