Universities Are Grading Your Process Now, Not Just Your Essay

University assessment is starting to focus less on the finished essay alone and more on how that essay was produced. Times Higher Education reports that 95 per cent of students now use AI in their studies, up from 66 per cent in 2024, while 12 per cent directly include AI generated text in assessed work, up from 3 per cent. That rise does not mean every use is misconduct. It does mean universities are looking for assessment methods that can separate legitimate AI support from work that cannot be confidently attributed to the student.

The clearest example is the two lane approach. From 2026-27, the University of Bath is replacing its traffic light system with open assessments, where AI use is optional or integral, and closed assessments, which are invigilated in person. Cardiff University has confirmed that it is implementing the same model and expects more emphasis on supervised formats where appropriate. In practice, this creates a sharper divide between tasks where AI can be part of the method and tasks where universities want direct evidence of unaided performance.

The shift is also visible in course design. The University of Surrey is redesigning its entire curriculum from September 2026 so that AI is embedded into every course. Rather than judging only the final output, Surrey plans to assess the process students follow and require them to evaluate AI results critically. That matters because the academic question is no longer simply whether AI was used. It is increasingly whether the student can explain the choices made, test the quality of the output and show where human judgement shaped the work.

One reason for this change is that simple permission rules are becoming harder to apply. Annika Bautz of the University of Surrey described traffic light systems as prohibitive, punitive and outdated. Andy Hamilton, a professor at Durham University, said he could no longer tell which students had used AI improperly and which had used it properly. If a marker cannot reliably infer AI use from the final prose, then the finished text alone becomes weaker evidence. A visible process, supported by drafts, notes and clear decisions, gives the assessment more context.

At the same time, universities are not abandoning traditional controls. A freedom of information study reported by Times Higher Education found that 78 per cent of UK universities still use online exams, only 10 per cent use remote invigilation for all exams, and an average of 246 examinations per university went unsupervised. Birkbeck, University of London reintroduced invigilated exams this academic year after a large number of academic misconduct allegations. The direction is mixed: more AI aware coursework on one side, and more supervised assessment where direct verification matters on the other.

Students are also being asked to navigate this shift before guidance is fully settled. The Digital Education Council AI in Higher Education Global Survey 2026 found that 88 per cent of students and 77 per cent of faculty use AI, yet 57 per cent of students say they lack adequate guidance for assessments. Only 29 per cent believe their instructors can guide them on AI use, while just 31 per cent of faculty say their institution involves them meaningfully in shaping AI policy. That gap makes documentation and careful reading of assessment rules more important, not less.

This term, a practical habit is to treat your working process as part of the assessment record. Keep your brief, notes, early drafts, source trail, prompts where permitted, and a short record of what you changed after using AI. Check the rules for each task instead of assuming one course policy applies everywhere. If AI is allowed, be ready to explain what it contributed and what you rejected. If it is not allowed, do not use it. Being able to show how the work developed is becoming a strong way to demonstrate authorship, judgement and academic integrity.