English regulator warns AI can undermine university assessments

The higher education regulator in England says students can complete some university assessments to a high standard using AI in minutes, raising concerns about whether those tests still measure students’ own knowledge and skills.

The Office for Students (OfS) said many assessments are unsupervised and AI use can be difficult to detect reliably. Its research, conducted with Advance HE, drew on views from 244 university staff and 91 students at at least 127 institutions. The findings describe risks reported by participants, rather than measuring how often students use AI across the sector.

The OfS said it had heard of cases in which students completed whole assessments in minutes using AI and received grades similar to those of students who did not. It also reported that some staff were seeing assignments that looked alike and questioned whether their institution’s assessments remained fit for purpose.

OfS strategy and delivery director Josh Fleming called for “far greater use of AI-free assessment”. He said AI posed a risk to students’ learning and universities’ ability to assess them. The regulator identified closed-book and practical exams as the most effective forms of assessment.

Universities UK, which represents more than 140 institutions, said many universities had made substantial changes, including more in-person exams, presentations, oral examinations and assessments that incorporate AI under controlled conditions. It said approaches should vary by subject.

The OfS says universities that fail to meet its quality standards could face restrictions on raising tuition fees or limits on student numbers under proposals now being consulted on. It also says that if an assessment can be completed using AI without demonstrating the skills it is meant to test, it should be replaced.