For years, digital exams were treated as a logistical upgrade. They promised fewer exam halls, faster administration, wider access and a cleaner way to manage assessment at scale. The appeal was obvious. Universities could run tests beyond campus walls, professional bodies could reach candidates in more locations, and students could sit exams without always travelling to a physical centre.
But the harder lesson is now difficult to ignore: digital exams are not simply a software problem. They are a trust problem.
A university can buy a stronger platform, add remote proctoring, tighten identity checks and lock down the browser. None of that automatically answers the deeper question students and academics are asking: does the exam still feel fair, valid and worth trusting?
Online Proctoring Made The Trust Problem Visible
Online proctoring did not create the trust problem in digital exams, but it made it much harder to ignore. Once the exam moved into bedrooms, kitchens, shared flats and family homes, assessment stopped being only an institutional process. It entered private space.
This matters because online proctoring does not simply move supervision from the exam hall to a screen. It can change the relationship between the student, the institution and the assessment environment. In a review titled Privacy as Contextual Integrity in Online Proctoring Systems in Higher Education, the authors frame privacy as more than discomfort with being watched. Their review of 17 papers shows that online proctoring can involve the collection and sharing of different types of personal and sensitive student information. The review also identified principles institutions need to reckon with if they want responsible adoption: transparency and fairness, consent and choice, information minimisation, accountability, information security and accuracy.
That gives the privacy concern more weight. The issue is not simply that students dislike being watched. It is that the rules of the exam can become unclear at the very moment the process becomes more intrusive. Students may not fully understand what is collected, who sees it, how long it is kept, what triggers a flag, or how much choice they realistically have. In that context, mistrust is not just emotional resistance. It is a predictable response to an assessment process that asks for compliance without always making the terms of that compliance clear.
Students experience those concerns while trying to sit still under a webcam, keep a quiet room, manage unstable internet, show identification and avoid behaviour that might be misread as suspicious. A system that asks students to prove they are alone, visible, quiet and technically prepared may protect exam integrity in one sense while weakening confidence in another.
Still, the argument should not stop at “online proctoring is bad”. The evidence is more complicated, and that complication matters. The ETS Research Report Detecting the Impact of Remote Proctored At-Home Testing Using Propensity Score Weighting examined whether at-home testing via remote proctoring created mode effects compared with on-site testing at test centres. Using data from three licensure tests, the study found that after adjusting for group differences with propensity score weighting, the estimated mode effects were small and nonsystematic overall, though there were some variations across subgroups based on gender and race.
That finding complicates the criticism in a useful way. It suggests remote proctored testing can produce results broadly comparable with test-centre delivery in some high-stakes contexts. But it also shows why “the scores are comparable” cannot be the only standard. Score comparability speaks to one kind of validity question. It does not settle whether students understood the process, felt fairly treated, had equal technical conditions, or trusted the institution’s handling of their data.
Better Software Cannot Fix Weak Assessment Design
The mistake many institutions make is treating digital assessment as if the main challenge is choosing the right tool. In reality, the harder challenge is designing the right conditions around the tool.
TEQSA’s guidance, Adapting assessment in the age of generative AI: The assessment adaptation model, matters because it moves the discussion beyond detection. Its focus is on adapting assessment design so tasks remain meaningful in a generative AI environment. That supports a wider point: digital risk cannot be answered only with tighter monitoring. It also requires assessment tasks that are harder to outsource, fake or reduce to a final submitted answer.
Generative AI has weakened old assumptions about take-home work, essays and online tasks. Universities can no longer assume that the final submission clearly represents a student’s independent thinking. The temptation is to respond with more control: locked browsers, webcam monitoring, room scans and stricter identity checks.
But security can become a narrow answer to a broad problem. An exam may be secure and still poorly aligned with the learning it claims to measure. A student may be prevented from cheating and still leave the process feeling that the institution cared more about suspicion than understanding.
Digital Delivery Exposes Old Problems
The most serious digital exam failures are rarely caused by one missing feature. They usually sit at the intersection of policy, pedagogy and operations.
Was the exam format suitable for online delivery? Were accessibility needs considered before the exam window opened? Were students told exactly what data would be collected and why? Was there a clear process for technical failure? Did staff know how to respond to exceptions? Were there alternatives for students who could not meet the home-environment requirements?
Software can support those decisions, but it cannot make them on behalf of the institution.
The study Hybrid-Flexible (HyFlex) Subject Delivery and Implications for Teaching Workload shows how easily institutions can underestimate redesign. The study found that adapting existing synchronous subjects to HyFlex delivery was not practically possible without significant redesign. Exams are no different. Moving an assessment online is not the same as redesigning it for an online environment.
That distinction is where digital exam strategy often becomes intellectually lazy. Institutions ask whether the platform can deliver the exam, but not always whether the exam should remain in its current form. They ask whether misconduct can be detected, but not whether the task itself invites shortcuts. They ask whether candidates can be monitored, but not whether the monitoring burden is proportionate to the stakes.
A Platform Is Infrastructure, Not A Strategy
Reliable platforms still matter. For large cohorts and high-stakes assessments, poor delivery can quickly become a fairness issue in its own right. The wrong system can create delays, technical failures, inconsistent candidate experiences and avoidable administrative pressure.
Dedicated assessment infrastructure has a legitimate role here. For example, an online assessment platform like janison.com can support secure test delivery, candidate management, monitoring, results processing and different assessment formats. However, better software should be treated as the floor, not the strategy. A platform can help manage delivery, security and administration. It cannot decide whether an exam has the right level of supervision, whether an open-book format would test judgement more effectively, or whether students understand the rules well enough to trust the process.
The Real Question Is What The Exam Is Trying To Prove
The next phase of digital assessment should be less obsessed with whether exams are online or offline, and more concerned with what the exam is trying to prove.
Some assessments will need strict supervision. Others may need authentic tasks, oral defence, practical demonstration, staged submissions, AI-permitted workflows or a combination of methods. The point is not to abandon exams. The point is to stop treating the exam format as sacred just because it is familiar.
Digital exams need better software, certainly. But they also need better institutional judgement. Without that, technology only makes old assumptions faster, larger and harder to question.
