AI-Exposed College Majors Linked to Weaker Employment and Pay, Study Finds

The rise of artificial intelligence (AI) coincides with a sharp deterioration in initial earnings and employment for graduates of the college majors most exposed to the technology, according to a new study.

Since the release of ChatGPT in late 2022, graduates in the 10 percent of majors most exposed to AI saw their likelihood of finding a job shortly after graduation fall by 5 percentage points, according to a working paper published by the U.S. Census Bureau’s Center for Economic Studies on Sept. 10.

Their initial quarterly earnings also fell by about 13 percent relative to graduates in fields less exposed to AI, according to the findings.

“This earnings decline is comparable in magnitude to the earnings losses associated with graduating into a large recession,” the researchers wrote.

The researchers analyzed administrative records covering roughly 29 percent of U.S. bachelor’s degrees awarded between 2016 and 2024. They found that graduates of the three majors most exposed to AI—computer science, computer and information systems, and computer engineering—experienced the largest declines in both employment and earnings after 2022.

“Fewer of these graduates found jobs immediately after graduation, and those who did so received lower earnings,” they wrote.

Graduates Shift Into Lower-Paying Work

According to the study, the decline in earnings among graduates from the most AI-exposed majors was driven by two main factors: industries that traditionally hired them began paying less, and more graduates took jobs in lower-paying industries.

“About half of the earnings decline reflects a shift away from high-paying industry sectors and toward lower-paying sectors such as Retail Trade and Accommodation and Food Services,” the authors wrote.

The earnings gap between graduates from the most and least AI-exposed majors narrowed over time, falling to about 5 percent two years after graduation.

Researchers also found higher job-switching rates, suggesting some graduates eventually moved into better-paying positions.

However, the authors said there is not yet enough data to determine whether those graduates will fully catch up with people who entered the labor market before generative AI became widespread.

Another Survey

In a recent survey of 261 graduates by the online learning platform Study.com, none of the respondents with computer science or information technology degrees said that AI made them wish they had chosen a different major.

By contrast, 43 percent of communications, journalism, and media graduates said they would choose a different major when considering AI’s impact. Biological sciences followed at 28 percent, and psychology at 22 percent.

Among all graduates surveyed, 77 percent said college was still the right choice despite AI’s impact on the job market, while 22 percent said they would have chosen a different major knowing what they know now about AI.

The findings are notably limited by a small sample size, as individual major groups included roughly one to two dozen respondents.

AI Is Not the Only Headwind

An earlier analysis from the Federal Reserve Bank of St. Louis suggests AI is only part of a broader problem facing young workers ages 18 to 24.

In a report published in June, economists at the bank found that a decline in job openings was the biggest contributor to weaker employment and higher unemployment among young workers from 2023 through 2025.

Growth in AI-related jobs also appeared to displace some new entrants, particularly recent college graduates, but its impact was smaller than the overall decline in labor demand.

Recent college graduates still had a 79 percent employment-to-population ratio nationally in early 2026, the St. Louis Fed noted, but a “low-hire, low-fire” labor market has made it harder for newcomers to get their first foothold.

“AI is not eliminating jobs across the economy. Instead, it is raising the bar for young workers trying to secure their first foothold in the labor market, an area in which employers feel less of a need to advertise openings, make offers and hire,” the authors wrote.