WEEKEND READING: Built for stability: quality assurance in an age of uncertainty

This blog was kindly authored by Dr Wai Mun Lim, Principal Researcher at the Institute for Adult Learning, Singapore University of Social Sciences.

There is a particular clarity that comes from distance. After 25 years inside UK higher education’s quality architecture, reviewing institutions for the QAA; serving on the subject benchmark panels; directing undergraduate programmes through to professional doctorate programmes; and being approved as an OfS reviewer before I left. I understood the system from the inside.  I believed in its purpose. I trusted its architecture.

Then, a year ago, I moved back to Singapore.

The distance made visible an assumption so deeply embedded in the system that I had never needed to name it. Quality assurance in UK higher education was built on the premise that the knowledge being assured is relatively stable, and that what a discipline contains; what a graduate should know; and what a curriculum should deliver can be specified, monitored, and verified over time.

Quality assurance works best when there is broad agreement on what constitutes knowledge in a field, what is considered rigorous, what a competent graduate should be able to demonstrate, and what a well-designed curriculum should include. Subject benchmark statements, learning outcomes, and external examining are mechanisms for verifying that institutional delivery aligns with a shared understanding of disciplinary content and standards. In this context, they serve as tools for maintaining consistency within a stable landscape.

For most of the system’s history, that stability was a reasonable working assumption. Disciplines had clear structures, and professional organisations understood their goals. Employers could specify the skills graduates should possess. Changes took place gradually, allowing review cycles to adjust. Although the system evolved, it did so within a framework where the key question of what knowledge is important was broadly answerable. This assumption is now being questioned because the mechanisms rely on it more than we previously realised.

Artificial intelligence is the most visible disruptor, but it is not the only one. In many sectors, especially outside technology, the relevance of specific bodies of knowledge is diminishing. What a graduate needs to know at the start of a career often differs from what will be needed in five years. In some fields, the very concept of disciplinary expertise is now uncertain in ways that were not present a generation ago. The question of what an economics graduate should know looks different in an era when AI can run models that previously required years of quantitative training.

This creates a particular challenge for quality assurance. Quality assurance can verify whether an institution delivers what it says it will. It is much less equipped to ask whether what is being delivered is the right thing, especially when that question no longer has a stable answer. When curriculum content, graduate outcomes, and employer expectations are all evolving at the same time, the tools designed to ensure consistency are essentially working against a moving target they were not designed to monitor.

The English settlement

In England, the apparent regulatory response to this pressure is to simplify rather than adapt. The Office for Students replaced QAA’s consensus-based approach with outcome metrics such as continuation and completion rates, graduate employment data, and, via the Teaching Excellence Framework, National Student Survey scores. These measures are straightforward and administratively convenient. Although OfS allow contextual mitigation where providers fall below thresholds, they are almost silent on whether what is being measured still aligns with what truly matters. The approach is aimed at reducing regulatory complexity and is not intended to nurture capacity for uncertainty. Therefore, the accountability framework will likely continue to report stability long after significant changes have occurred, with no mechanism to inform us otherwise.

The challenge goes deeper than implementation, as it is not that quality assurance systems are poorly designed or managed. It’s that they are being asked to assure something they were never meant to achieve, even as reviews are carried out diligently, panels contribute expertise, and institutions take the process seriously. The difficulty runs deeper because the architecture assumes that the right questions can be defined beforehand and that progress towards answering them can be tracked. When the questions themselves remain unresolved, that assumption quietly falls apart.

Challenged with this level of structural limitation, the temptation is to update the framework, add new criteria, broaden the definition of outcomes, and commission guidance on AI readiness. These are reasonable responses, and they are already happening. But they risk mistaking a design question for an implementation problem.

Quality assurance can be refined, expanded, and made more responsive. What it cannot do is assure quality in domains where quality itself is genuinely contested or unknown. When we do not yet know what excellent AI-era curriculum design looks like, or which graduate capabilities will matter most in five years, asking institutions to demonstrate that they are delivering it, and asking reviewers to verify that they are doing so, places both parties in an impossible position.

Naming the limit

The honest assessment is that quality assurance does some things well and will continue to do so. Baseline standards, institutional accountability, and consumer protection for students remain valuable and achievable. But the capabilities that matter most for weathering an uncertain future may lie beyond what any external verification system can reliably assess. That gap deserves urgent attention, not necessarily just better documentation.

Identifying this limit is the easy part. Human capabilities that are critical during uncertain times, such as working under ambiguity, continuous learning, and crossing disciplinary boundaries, require institutional conditions that external verification often overlooks. These include pedagogy, culture, professional judgement, and the quality of the relationship between educators and learners. While these are consistent and significant, they remain largely invisible within the accountability structures we have established.

While this does not mean we abandon accountability, it suggests examining what accountability can realistically achieve and, alongside that, developing ways to address its limitations. Other fields have faced similar issues; for example, medicine distinguishes between measurable clinical outcomes and the more subjective qualities of good clinical judgement, as reflected in the GMC’s Good Medical Practice framework (2024). This framework separates specific performance standards from the professional judgement doctors must exercise when applying them. Clarifying what reassessment involves may require a broader discussion. Nonetheless, it will likely start with assessing which functions quality assurance can perform effectively and with dedicating separate efforts to the institutional conditions, teaching cultures, and professional communities of practice that accountability frameworks cannot fully address. I left the UK quality assurancesystem with respect for its purpose. Now, from a distance, I revisit it and wonder if we have built an architecture so focused on what can be verified that we have quietly stopped asking whether that verification still asks the right question?