42% of College Students Say AI Changed How They Think About Their Major

What Students Are Thinking, and Why

The study does not only measure the incidence of AI-influenced major changes: it provides qualitative context for why students are making these reconsiderations. The most common concern, cited by 67 percent of students who reported AI influencing their major thinking, was a belief that the career they had been preparing for would have "significantly fewer job openings" within 10 years than they expected when they enrolled. That perception is not uniformly accurate, and the study acknowledges that students may be overestimating AI's near-term displacement impact in some fields, but the perception itself is real and is shaping enrollment decisions regardless of whether the underlying labor market fear is proportionate.

The students most likely to have changed majors specifically because of AI are concentrated in communication, journalism, paralegal studies, basic finance and accounting, and some areas of design and creative work where AI-generated outputs have become visibly competitive with entry-level human work product. These fields share a pattern: they require skills that are clearly articulable, that produce outputs with measurable quality dimensions, and that AI training data has been able to capture at a level of competence that overlaps with the entry-level range of the profession.

"Students are not reacting to abstract fear of AI. They are watching what is happening to people who graduated two or three years ahead of them, and they are updating their plans based on what they see. That is rational behavior. What it tells us is that higher education needs to update the value proposition it is offering, not just reassure students that their fears are overblown."

Stacy Pryor, Vice President for Learning and Economic Mobility at Lumina Foundation, in remarks to the Washington Post, March 2026

What Fields Students Are Moving Toward

The study pairs its findings on AI-related major changes with data on where students are redirecting: the fields gaining enrollment share from students who have shifted away from AI-vulnerable programs. The patterns are consistent across institution type and geography, and they reflect a student understanding of AI's impact that is more nuanced than simple technology fear.

Healthcare professional programs, particularly nursing, physical therapy, occupational therapy, and physician assistant studies, are the single largest category of enrollment growth driven by AI-adjacent reconsiderations. Students articulate a clear rationale: these professions require hands-on patient contact, physical dexterity, and contextual clinical judgment that current AI systems cannot replicate in the patient-facing dimensions of the work. The credentials also have clear licensing structures that create professional entry barriers not present in many other fields.

Trades and technical credentials, accessed through community college and apprenticeship pipelines, are gaining from students who might previously have pursued four-year communication, business administration, or liberal arts degrees. The convergence of strong trades labor market conditions, visible AI displacement of some office work, and growing social recognition that trades skills have been undervalued relative to their actual economic contribution is driving a demographic that would not previously have considered a trades pathway to evaluate it seriously.

Computer science and data science programs are the third major growth category, and the motivation is different from the defensive calculations driving healthcare and trades growth. These students are moving toward AI-proximate fields not because they think those fields are immune to AI displacement but because they want to be on the side of the technology rather than subject to its impact. The enrollment growth in these programs is happening at the undergraduate level even at institutions where the programs were already oversubscribed, and it is forcing curriculum and faculty expansion questions that departments were not previously planning to address at this speed.

By Institution Type: Four-Year vs. Community College

The 14-percentage-point gap between four-year students (42 percent influenced by AI) and two-year students (56 percent reconsidering) reflects real differences in the economic context of the decision. Community college students are overwhelmingly pursuing education as a direct pathway to employment, and the time between enrollment and expected graduation is typically two years or less. That short horizon makes the perceived accuracy of their career forecast more consequential: if they are wrong about where entry-level hiring will be in two years, the cost of the error arrives quickly and concretely.

Four-year students have a somewhat longer runway between their current enrollment decisions and their labor market entry, which may explain the lower rate of stated AI influence: with four years between now and job-seeking, the AI labor market landscape is harder to forecast accurately enough to justify a major change. The 16 percent actual major change rate among four-year students, however, suggests that for those in the middle of their programs who have a clearer line of sight to their graduation timeline, the perceived need to act is real.

First-generation students, those who are the first in their family to pursue a college degree, show higher rates of both AI influence on major thinking (51 percent) and actual major changes (21 percent). The Lumina research suggests this reflects first-generation students' closer attention to the practical labor market value of their education: students who cannot rely on family wealth as a fallback are more sensitive to evidence that their credential pathway may not deliver the employment outcomes they need it to deliver.

The Institutional Response Problem

The 42 percent figure creates an institutional challenge for colleges and universities that is distinct from the student decision challenge. Higher education institutions have historically operated on curriculum revision cycles that measure their pace in years to decades: a new major program typically takes three to five years from faculty proposal to accreditation to student enrollment. That timeline is incompatible with the pace of AI labor market change that students are observing and responding to.

Several institutions have begun creating faster-track credential pathways, micro-credentials, and certificate programs that can be added to existing degree pathways without requiring the full curriculum revision process. The Lumina Foundation's framing of the study explicitly advocates for this approach, describing it as "building AI fluency into existing programs faster than new programs can be created."

The study's most pointed institutional finding is that the students who felt their institution was actively helping them understand AI's impact on their career field were significantly less likely to have changed majors out of AI-related concern: they were making informed decisions to stay rather than making uninformed decisions to leave. That difference in experience, between students who felt guided through the AI-employment question and those who felt left to figure it out themselves, is the gap that the most responsive institutions are working to close.

The AI skills gap across the broader workforce that other research has documented is in some sense a downstream version of the problem this survey captures: students who are uncertain about how to position themselves in an AI-affected labor market become workers who are uncertain about the same question, and the institutional responsibility to provide clarity runs from current enrollment decisions through to the retraining and continuing education programs that employed workers will need as the market continues to evolve.

What the Data Means for the 2026 Enrollment Cycle

The Lumina-Gallup findings arrive just before the spring enrollment decision season, when prospective students are making final choices about where to enroll and in what programs. The study's 42 percent figure is likely to amplify existing trends: students who were already considering healthcare, trades, and technology pivots will feel more validated; institutional admissions teams are going to face more questions about career placement and AI-proofing at their information sessions; and program directors in communication, business administration, and liberal arts departments will have new conversations to navigate.

The most important contextual note for students interpreting this data is that the study measures fear and current behavior, not optimized career strategy. The fact that 42 percent of students report AI influencing their major choice does not mean that 42 percent of students should change their majors, or that the majors they are considering changing from are necessarily poor choices. The relationship between current AI capabilities, future labor market outcomes, and the career value of specific credentials is genuinely uncertain, and students who are making major changes based primarily on generalized AI anxiety may be responding to noise rather than signal.

What the data does definitively establish is that students are paying close attention to the relationship between their education and their employment prospects, that they are not assuming this relationship is automatic or guaranteed, and that they are willing to make consequential decisions based on their best understanding of where the labor market is heading. That is exactly the kind of informed consumer behavior that higher education institutions say they want from students. The harder question is whether institutions are ready to respond to it with the speed that students' decision-making timelines require.