Universities are redesigning assessment at speed as generative AI moves from the margins of student life into the center of how coursework is written, checked and graded. Recent reporting shows institutions in the UK, Europe and the US shifting away from take-home essays and toward supervised exams, oral defenses and “closed” assignments that are harder to complete with AI assistance.Nature Times Higher Education
The change reflects a practical problem as much as an academic one. In a 2026 survey of 1,054 UK undergraduates cited by Nature, about 94% said they use generative-AI tools to help with assessed work, while 12% said they had directly inserted AI-generated text into coursework.Nature Educators are also reporting suspicious markers in submissions and examinations, from fabricated references to abrupt changes in style, which has intensified debate over how to preserve academic standards without criminalizing routine use of new tools.Nature
Assessment is being rebuilt around process
Some universities are responding by putting more weight on the learning process rather than the final text on the page. The University of Bath, for example, has said that from the 2026-27 academic year it will move to a simpler “two-lane” assessment model that separates open tasks, where generative AI may be optional or even integral, from closed, invigilated exams.Times Higher Education
That approach mirrors a broader trend. Reporting from AP and other outlets describes colleges turning to in-person tests and oral exams to counter AI-enabled cheating, including AI-powered oral assessments in business and management courses.AP News Some professors argue the shift restores accountability and gives instructors a clearer view of what a student can explain in real time, rather than what an algorithm can draft in minutes.Los Angeles Times
But the move back to supervised testing is not universally welcomed. Students and some faculty members say more in-person testing can increase stress, favor those who perform well under pressure and penalize students who balance study with work, caregiving or disability-related needs. A push toward tighter surveillance also risks turning assessment into an arms race, especially as wearable AI tools become harder to police.Times Higher Education
“The transformation, effective from September 2026, will see it move towards assessing students’ process over their outputs,” Times Higher Education reported of Bath’s new model.Times Higher Education
The job market is pulling universities in the same direction
The assessment debate is also being shaped by what employers say they need. Universities are under pressure to show that graduates can communicate, defend decisions and solve problems in settings where AI is a common workplace tool. That makes oral exams, project presentations and supervised tasks attractive to institutions trying to prove degree value in a labor market that is increasingly skeptical of paper credentials alone.Times Higher Education
At the same time, early-career hiring remains active but highly structured, with employers advertising internships and graduate roles that often require students to signal readiness long before graduation. That reality strengthens the case for “work-like” assessments that mirror the expectations of entry-level jobs: clear writing, live explanation, collaboration and the ability to use AI without becoming dependent on it.Isio Extern
Career-services advisers are already telling students that the transition from university to work depends on more than grades. Penn Career Services’ post-graduation reporting points to the importance of internships, networking and practical experience in landing the first role after college, a pattern that universities are now trying to reflect more explicitly in their curricula.Penn Career Services
Students want clarity, not just crackdowns
For students, the central complaint is often not that universities are regulating AI, but that the rules change too fast or vary too widely between courses. A tool that is permitted in one seminar may be forbidden in another, and AI-detection systems remain controversial because of false positives, especially for non-native English speakers and other students whose writing style may not match a detector’s expectations.Study International
That uncertainty is pushing some institutions to focus less on surveillance and more on clarity. Clear instructions about when AI may be used, how sources must be disclosed and what counts as original work are becoming essential as universities try to balance academic integrity with the reality that many students now use AI as a brainstorming, editing or tutoring tool.Nature Study International
The broader lesson is that higher education is no longer deciding whether AI belongs in the classroom. It is deciding how to assess students in a world where AI is already there, and where the answer will shape not only academic honesty but the kind of graduates employers receive.
Sources: Nature; Times Higher Education; AP News; Los Angeles Times; Study International; Isio; Extern; Penn Career Services