47% of US College Students Reconsidered Their Major Because of AI
According to Gallup’s April 2026 account of the survey, 47% of US associate- and bachelor’s-degree students had given at least a fair amount of thought to changing their major because of AI’s potential impact, while 16% said they had already changed fields. The finding remains a significant measure of student concern, but it describes self-reported deliberation and decisions rather than a verified nationwide transfer between particular programs.
Evidence published after the survey adds movement but not a simple verdict. Computer-related enrollment fell in spring 2026, even as total undergraduate enrollment grew, while the latest federal projections still anticipate strong expansion in several occupations associated with developing, securing and applying AI. Together, the updates make it harder to argue that students are moving toward one clearly safer academic path.
What the headline figure measures
The underlying web survey was conducted from October 2 through October 31, 2025, among 3,801 US students aged 18 to 59 who were pursuing associate or bachelor’s degrees. Respondents came from an opt-in online panel, so the result should be read as a survey estimate for the defined student population, not as an administrative count of every major change made at US colleges.
“Reconsidered” also covers a wider range of behavior than formally switching programs. A student who seriously weighed another field but remained in the original program belongs in the first group, while the smaller group who said they had changed fields reported a completed decision. The study did not establish that AI was the only cause of those choices, nor did it provide a complete map of the majors students left and entered.
That distinction matters because changing a major can reflect several overlapping pressures. Expectations about employment may interact with tuition costs, time to graduation, academic performance, access to required courses and a student’s changing interests. The survey establishes that respondents connected AI to their thinking; it does not isolate the technology from every other influence.
Enrollment shifted after the survey, but causation remains unproven
The National Student Clearinghouse’s June 2026 report found that spring undergraduate enrollment increased 1.3% year over year to 15.5 million, while Computer and Information Science enrollment declined 8.4% at undergraduate four-year institutions and 11.2% at two-year institutions. This is a concrete change in program enrollment after the attitudes captured by the earlier survey.
It is not, however, a direct count of students who switched because of AI. Enrollment totals combine incoming, continuing and returning students, whereas the survey asked enrolled students to describe how AI had affected their own decisions. A decline can therefore reflect fewer new entrants, transfers, program reclassification or departures from college as well as changes of major.
The timing alone cannot establish a causal link. Computer-related programs had previously experienced rapid growth, and enrollment can respond to institutional capacity, demographics, local availability, perceived hiring conditions and the cost of completing a credential. The defensible conclusion is narrower: anxiety about AI and a decline in computer-related enrollment appeared during the same broad period, but the available evidence does not show how much of the decline AI caused.
The employment outlook does not support a single “AI-proof” major
The US Bureau of Labor Statistics projections published July 16, 2026 anticipate employment growth from 2024 to 2034 of 15.8% for software developers, 33.5% for data scientists, 28.5% for information-security analysts and 19.7% for computer and information research scientists. The same federal outlook expects weaker demand in several office and administrative occupations where automation may produce efficiency gains.
These contrasting directions show why “technology jobs” and “jobs threatened by technology” are not mutually exclusive categories. AI can reduce demand for particular routine tasks while increasing demand for workers who build systems, protect information, analyze data or apply technical tools in specialized settings. Exposure to AI is therefore not equivalent to the disappearance of an occupation.
Projections also describe employment across the economy, not guaranteed openings for a new graduate in a particular city or year. They include experienced workers and cannot capture every future change in technology, investment or hiring. Conversely, a difficult entry-level market at one moment does not by itself establish that an occupation will contract over the full projection period.
A major is broader than a single occupational forecast. Graduates from one program can enter several kinds of work, while employers in software, data and AI-related teams may recruit people with different academic backgrounds. The relationship between degree title and career outcome is therefore too loose to rank majors by one measure of automation risk.
What the evidence supports
AI is already influencing academic decisions, but the available data do not identify a permanently protected field. The survey shows that concern has moved beyond abstract discussion for a substantial share of students. The later enrollment figures show a pronounced retreat in one broad academic category, but they do not reveal the motives behind every enrollment decision.
The labor projections add the central complication: some occupations closely associated with AI are expected to expand rapidly, even as automation restrains demand elsewhere. That makes a broad flight from technical study no more evidence-based than assuming every computing credential will guarantee employment.
The clearest reading is that students are responding to genuine uncertainty rather than to a settled map of winners and losers. AI may change the tasks performed within many professions before it eliminates or creates entire occupational categories. For students evaluating a major, the consequential question is therefore not whether a field carries an “AI-resistant” label, but how its curriculum develops subject knowledge, judgment and responsibility alongside changing tools.
Also read:
Subscribe to our newsletter
Get the latest Web3, AI, and crypto news delivered straight to your inbox.