For students choosing degrees and preparing for their first jobs, artificial intelligence is no longer a distant workplace debate. It is becoming a practical question: which tasks will remain valuable, and how should universities prepare people for careers that may change faster than their courses?
The answer is still unsettled. Recent evidence does not show a broad collapse in employment for university graduates, but it does point to sharper pressure in some fields, particularly technology and other occupations exposed to generative AI.
A policy brief from Stanford’s Institute for Economic Policy Research said unemployment among recent graduates reached 5.6% in early 2026, 1.6 percentage points higher than three years earlier. The figure has intensified concern that AI is reducing opportunities traditionally used as entry points into professional work. Stanford SIEPR
The pressure is concentrated, not universal
Research from the Federal Reserve Bank of Dallas found that recent graduates in more AI-exposed majors at Texas universities experienced a 1.7-percentage-point decline in the probability of finding employment in Texas within a year of graduation, relative to graduates in less-exposed majors. Those graduates were also more likely to return to education, suggesting that further study can become a response to a weak first-job market rather than a purely academic choice. Federal Reserve Bank of Dallas
The findings do not mean that a computer science or other technical degree has lost its value. They suggest instead that the transition from university to work may be changing. Employers can use AI to complete some junior-level writing, coding, research and administrative tasks, potentially reducing the number of conventional starter roles while increasing demand for workers who can supervise systems, evaluate outputs and apply technical knowledge to specific industries.
Other sectors remain more resilient. An analysis published by USA Today reported that health, education and social services were hiring at close to normal levels, while entry-level recruitment had slowed most noticeably in technology-related areas. It also cited data indicating that 77% of graduates from the class of 2025 found jobs within three months of receiving their degrees. USA Today
Universities face a difficult redesign
Higher education institutions are responding with new AI courses, credentials and policies, but many are still deciding what responsible use should look like. UNESCO reported that two-thirds of surveyed higher education institutions either had guidance on generative AI or were developing it. The organization has also published competency frameworks for students and teachers focused on using AI effectively and responsibly. UNESCO
Policy, however, is only one part of the challenge. Universities must decide whether AI belongs in assessment, how students can use it without outsourcing their learning, and which traditional assignments still measure the skills employers need. Replacing essays with oral examinations, project work or supervised practical tasks may improve confidence in assessment, but such changes can increase staff workloads and create accessibility concerns.
There is also a risk of treating AI literacy as a substitute for education. Students need to understand tools, but they also need subject knowledge, judgment, communication and the ability to work with people. Those foundations help graduates identify inaccurate outputs, protect confidential information and make decisions when no automated answer is reliable.
Students are being asked to prepare for uncertainty
That uncertainty creates unequal burdens. Students with access to internships, professional networks and high-quality AI tools may be better positioned to experiment and build portfolios. Others may graduate with less practical experience while competing for fewer entry-level roles. Universities that expand AI training without addressing work placements, careers advice and affordability could deepen those differences.
The most credible response is neither to dismiss AI as hype nor to present it as an unavoidable replacement for human workers. Degrees still provide disciplinary knowledge and a route into regulated professions, while the employment payoff varies by subject, location and economic conditions. But students may increasingly need evidence of applied ability alongside a qualification: projects completed with real organisations, experience checking automated work and examples of sound judgment.
The future of graduate employment will therefore depend on more than how quickly universities add AI modules. It will also depend on whether employers continue to create genuine entry-level pathways and whether institutions can make practical experience available to students who cannot afford unpaid opportunities. For the next cohort, the central question is not whether AI will change work. It is whether education and hiring systems will change quickly enough to ensure students benefit from that transformation.