There is a sentence in coverage of new sector research on AI in learning and teaching that stopped me this week. Reporting on a series of round tables with senior education leaders, the authors noted that traditional mechanisms for managing and developing learning and teaching may be ill-matched to the pace of change that AI represents. It is stated almost as a given. But it seems to me to deserve some further examination.
The observation is structural, not technical. It is not about whether AI tools are useful, or whether they threaten academic integrity, or what graduate employers will eventually expect of AI-literate graduates. It is about the institutional architecture through which curriculum and pedagogical change is governed, approved and embedded. And that architecture, in most UK universities, was designed for a different set of conditions.
Consider what the standard mechanisms look like in practice. Annual monitoring reviews curriculum performance against agreed metrics. Module approval panels scrutinise changes to learning outcomes, assessment and content. Periodic review subjects whole programmes to scrutiny every five or six years. Validation processes govern the introduction of new provision. Staff development frameworks operate on rolling CPD cycles, often with annual appraisal as the primary lever. These mechanisms are not arbitrary bureaucratic obstacles. They exist because quality assurance, comparability and academic standards matter. They were designed to manage incremental drift, to catch problems before they compound, to maintain coherence across a programme over time.
The difficulty is that AI may be something qualitatively different from incremental drift.
In some areas of AI capability, relevant use-cases and tool functionality are shifting on a less than twelve-month cycle. A pedagogical response developed to address the situation as it stood in autumn 2024 may already require revision. The institution that managed to get a modified assessment strategy through its approval processes by spring 2025 may find itself revisiting that work before the next academic year is out. The temporal mismatch between the rate at which AI is changing and the rhythms at which institutional review processes operate is not merely inconvenient. It may be structurally significant.
This is not a criticism of the mechanisms themselves. It is an observation about the conditions under which they were designed, and whether those conditions still hold. The question worth asking is whether any approval process built for gradual, manageable change can accommodate change that is neither gradual nor obviously manageable.
The discipline dimension deserves particular attention. The research notes that leaders are working to develop discipline-appropriate pedagogical responses, working alongside educators in their subject communities. I think this is entirely sound. But discipline-level change in higher education is mediated through some of the slowest-moving structures in the system. Programme teams operate within institutional frameworks. Subject benchmark statements, maintained by the Quality Assurance Agency, are revised periodically but not rapidly. Professional body accreditation requirements, which govern provision in fields from nursing to law to architecture, carry their own review timescales and their own accountability obligations. None of these structures are obstacles in any simple sense. Each exists to protect something that matters. But each also has a rhythm, and those rhythms may not readily accommodate what is being asked of them.
Sitting between the pace of external change and the pace of internal process are educators. The research acknowledges honestly that some hold fundamental criticisms of AI and that institutions are, up to a point, accepting this. That is a reasonable and humane acknowledgement of a real situation in which thoughtful professionals hold principled positions. But it also suggests that the current period may be quietly producing a widening divergence: between those who are actively integrating AI into their practice, those who are cautiously experimenting and those who are exercising what they regard as principled restraint. It is not obvious that existing staff development frameworks are well-configured to manage that divergence constructively, let alone to support all three groups in ways that are educationally coherent and institutionally consistent.
The research finds that institutional leaders are, on the whole, at an exploratory stage. That seems honest and appropriate. Given the genuine unknowns about how AI will manifest in secondary education, what employer expectations will settle into, which tools will survive market consolidation etc. so it would be surprising if most leaders claimed otherwise. Caution makes sense.
But there is a version of caution that is genuinely exploratory, actively building the institutional capacity to respond as conditions become clearer. And there is a version that is exploratory in a more passive sense, in which the exploration is happening at the level of individual educators and discipline communities, whilst the infrastructure through which their findings would eventually be formalised and embedded remains largely unchanged. The question of which version is more common is worth sitting with.
It is possible that the answer lies not in abandoning existing mechanisms but in developing supplementary structures that can operate at a different pace. Some institutions appear to be creating dedicated AI roles, task forces or curriculum development units intended to work with greater agility. Whether those structures have the authority and resource to translate their work into lasting changes in curriculum and assessment practice, rather than simply advising on them, is a separate and important question.
What would institutional development infrastructure need to look like if it were designed not for incremental change but for conditions of genuine uncertainty and rapid external movement? Who in the institution has both the authority and the remit to work outside normal approval timescales when those timescales prove genuinely inadequate? And are institutions being sufficiently honest with themselves about the gap between what they are asking of educators and the support they are in a position to offer? Those questions may not have ready answers. But everything we are hearing suggests they are no longer possible to defer.