Secure the Assessment Without Locking Out the Learner

Universities are entering a new phase of the artificial-intelligence debate. The question is no longer only whether students may use generative AI. It is how an institution can honestly certify that a graduate possesses the knowledge, judgement and capabilities represented by a qualification.

Reporting published on 18 July describes the Australian National University consulting on whether assessments should be classified as secure or insecure. Other Australian universities are expanding supervised examinations and oral assessment. The proposals have also drawn criticism from educators concerned that a hurried return to uniform, high-stakes conditions may disadvantage students with disabilities, caring responsibilities or other legitimate access needs. ANU says no final decisions have been made and that consultation will continue. The tension is real: institutions must protect the credibility of qualifications without making educational inclusion collateral damage.

A secure examination can show that a particular answer was produced under controlled conditions. It does not automatically show that the examination measures the most important outcome of the programme. A student may recall information in a supervised room yet remain unable to interpret a difficult case, exercise ethical judgement, communicate with people or transfer learning into practice.

The better question is not simply, “How do we stop unauthorised AI use?” It is, “What evidence would justify our claim that this learner is competent?” Security is one possible condition for gathering that…