A study published this month in the journal Frontiers in Education argues that universities have misdiagnosed their AI problem, spending more energy catching students who use chatbots than preparing them for a job market where AI use is routine.
Beyond plagiarism detection
The paper, by Dr. Kelechi Ekuma of the University of Manchester’s Global Development Institute, examines development studies programs specifically but frames its conclusions as relevant across higher education broadly. According to Ekuma, “The debate about AI in universities has often focused on whether students are using chatbots to complete assignments. While those concerns are understandable, they risk missing a much bigger transformation.”
The study contends that plagiarism-detection software and blanket bans treat AI mainly as something to police, rather than as a condition graduates will work within for the rest of their careers. It proposes shifting assessment away from polished written output — the kind a chatbot can produce easily — toward formats that are harder to outsource: oral examinations, reflective accounts of how AI was used during an assignment, collaborative projects and real-world problem-solving exercises.
Five capabilities, not one course
Rather than isolating AI in a single specialist elective, the paper argues AI literacy should be threaded through existing modules on governance, inequality, labor markets and research methods. It identifies five capabilities universities should build: understanding what AI systems can and cannot do, recognizing bias and hallucination, exercising complex judgment, applying ethical analysis, and adapting to changing tools.
Ekuma also argues that graduate employability itself needs redefining — from technical proficiency in using a given tool toward what the paper calls “adaptive capability”: the ability to interpret, question and govern AI-mediated systems rather than simply operate them.
The paper lands as universities worldwide continue to search for a coherent AI policy, several years after chatbots first upended coursework and exams. Ekuma’s argument is that the search has focused on the wrong question — not how to catch AI use, but how to teach around it.