Which Jobs Are Most at Risk in the Age of AI?

In the three-plus years since large language models went mainstream, college students have been inundated with the tech sector’s gloomy predictions that artificial intelligence is coming for their jobs. And so far in 2026, those predictions have only become more extreme.

In February, Microsoft’s AI chief declared that all white-collar work would be automated within 18 months. Soon after, Anthropic’s CEO doubled down on earlier assertions that AI could wipe out half of all entry-level white-collar jobs by the end of the decade, describing this moment as humanity’s “rite of passage.”

But big tech companies aren’t the only ones analyzing and forecasting how the widespread adoption of AI-powered products is reshaping the labor market.

Last month, researchers at Tufts University published “When Wired Belts Become the New Rust Belts: AI and the Emerging Geography of American Job Risk,” which ranks occupations, industries, regions and states by vulnerability “based on the most current understanding of AI’s evolving impact.”

While not as severe as the tech sector’s predictions, the index projects that roughly 6 percent of jobs are vulnerable to AI-driven elimination within the next two to five years, amounting to “a wipeout equivalent to the economy of Belgium” or even “just shy of the economy of South Korea,” if adoption of agentic AI tools increases.

Some workers—and prospective workers—should be more worried than others, the report says.

According to the report, the information, finance and insurance, and professional, scientific and technical services sectors are most vulnerable, with a quarter of job losses expected to come from just eight occupations. The most vulnerable include writers and authors, computer programmers and web and digital interface designers, who all face job losses of more than 50 percent. Meanwhile, 38 percent of jobs are still considered AI-proof. However, many of those are lower-paying and don’t require a college degree—such as roofers, school bus drivers and medical assistants—putting “the safe zone” at the “near-poverty zone,” noted the report.

The report also projects that major metro areas and college towns will face the highest rates of displacement, with four in 10 AI-related job losses located in California, Florida, Illinois, New York and Texas.

“AI-driven job vulnerability is uneven but material,” the authors of the report wrote. “Even as the technology continues to evolve—with breakthroughs and setbacks alike—and as organizations and workers adapt in real time, the broad outlines of the emerging geography of American job risk due to AI are becoming clear.”

Although the report’s projections about AI-related job displacements offer some new insight, it builds on a growing body of academic research. Over the past year or so, researchers at Yale, Stanford and the Massachusetts Institute of Technology have published reports about which career fields are most exposed to AI-driven automation or augmentation.

So far, jobs related to writing and coding—among others that often require a college degree—have consistently ranked highest.

No matter how alarming or disruptive these findings may be to higher education institutions preparing students for the workforce, job-market experts say colleges and universities can’t afford to ignore these emerging projections about AI.

“Job loss is going to happen,” said Gad Levanon, chief economist of the Burning Glass Institute, a nonprofit research group focused on the future of work. “I wouldn’t be surprised if we are at the beginning of decades of job displacement caused by AI.”

Instead of avoiding or minimizing the issue, “universities should acknowledge that things are changing very rapidly and do the best they can to prepare their students for the new labor market.”

Although the data about AI-related job-loss projections could always be more nuanced, Tiffany Hsieh, a senior director in the Center for Artificial Intelligence and the Future of Work at Jobs for the Future, said the information in the Tufts report and others can inform institutional priorities or changes in response to the integration of AI.

“We have enough of a sense from the existing body of research that there is a disruption coming,” she said. “We’re starting to see an alignment on where the occupational impacts will be, and we need to act now because our systems aren’t set up to move very quickly. [Higher education] needs to think about what we can do now to fuel the changes that need to happen when this disruption actually comes. We don’t want to be caught flat-footed.”