BUSINESS REFLECTION: Crossed Wires: Degrees of doubt — how generative AI is fracturing higher education

The world of higher education is facing an unprecedented crisis due to the rise of generative artificial intelligence (AI). Universities are struggling to assess learning and risk reducing formal degrees to mere certificates of attendance. In the past, these institutions put their reputation on the line by guaranteeing that students had learned something, but now they cannot make that claim with certainty.

Generative AI is shattering university testing standards, leaving institutions unsure of what they are certifying. An essay on the causes of World War 1, for example, could be written by a chatbot in 40 seconds, and the marker would not be able to prove whether the student actually wrote it or not. This uncertainty has led to a breakdown in the trust that universities have traditionally relied upon when awarding degrees.

The numbers are alarming. A survey by Hepi, a British higher education think-tank, found that 92% of undergraduates now use generative AI in some form, and 88% use it in their assessed work. This is despite the fact that AI models are becoming increasingly capable, with better research, reasoning, problem-solving, and more natural prose.

Regulators are struggling to keep up with the speed of AI's progress. Research from previous years is becoming useless, and current studies of student behavior are constrained by the cadence of scientific publishing. This makes it difficult to collect and distribute evidence in a timely manner, which further complicates the issue.

Tools designed to catch cheaters are also problematic because they have a high rate of false positives. An independent test of 14 tools found none exceeding 80% accuracy. No one wants to falsely accuse a significant portion of the student body, as this would be a legal and ethical nightmare.

As a result, universities are abandoning detection tools altogether. The University of Cape Town, for example, switched its detector off in October 2025. Other institutions like Waterloo, Curtin, Vanderbilt, and Washington State have followed suit.

Technical workarounds are being developed to circumvent these detection tools, making the problem even more complex. Proscription, sanctions, bans, and public shaming are blunt instruments that simply won't work in this situation.

The situation is further exacerbated by the sheer number of AI-written assignments. Universities are noticing the epidemic proportions of AI usage, and regulators have conceded that some or all of the assessments contributing to the evidence base "may no longer be confidently ascribed to an individual student".

The crisis is compounded by the fact that many students are not adequately prepared to use AI. A February 2026 study found that 86% of students reported using AI, but only 23% felt properly prepared to use it. Most students cited inadequate institutional guidance.

In response to this crisis, universities are experimenting with oral examinations and multiple-choice exams. However, these methods have their own drawbacks. Oral examinations are time-consuming, expensive, and do not scale well, while multiple-choice exams can only probe certain types of knowledge.

The lack of a uniform approach to AI usage and guidance has left students and institutions alike confused and uncertain about what they are learning and how to assess it. With no clear consensus on how to move forward, the future of higher education in the age of generative AI remains uncertain.