Jisc's 'human in the loop' pilot sharpens AI governance for student feedback evidence
Updated Jun 20, 2026
Jisc's latest AI proposal is not really about buying another tool. It is about turning "human in the loop" from a reassuring phrase into a governed institutional process. On 18 June 2026, Jisc published What does “human in the loop” actually mean? Consulting on our next pilot idea, proposing a new 2026-27 pilot on meaningful human oversight in AI-enabled workflows. For institutions that collect and act on student voice, that matters because AI-supported feedback and AI-assisted evidence are now close enough to practice that governance gaps will start showing up in student comments, committee papers, and quality reviews.
What the "human in the loop" proposal changes
The immediate shift is strategic. Jisc says earlier pilots made sense when institutions needed access to tools and wanted help deciding whether products were useful. The new question is different: many colleges and universities already hold licences for relevant tools, so the issue has moved from product choice to governed use. Jisc says institutions are now asking not "is this tool any good?" but "how do we use AI well?" That is a more demanding question, because it pushes universities towards policy, review processes, staff capability, and evidence of oversight rather than tool discovery alone.
"The tool is the medium; the practice is the subject."
Jisc's proposed model has two phases. Phase A is a structured discovery programme for member colleges and universities, designed to co-produce formal rules, guidance, ethical checklists, review processes, and evaluation approaches. Jisc says that stage should be a genuine exit point, so an institution that stops there still leaves with useful governance materials. Phase B is optional and would test those outputs in live settings, using AI-enabled tools institutions already have access to. The aim is a practical framework for meaningful human oversight that can be refined through use and then published for the wider membership.
The scope is broader than assessment alone, even if assessment and feedback are the clearest examples in the article. Jisc says the pilot is open to all member colleges and universities and is intended for anyone whose work involves AI supporting judgements or content that a human is expected to oversee, including educators, assessment leads, quality and governance teams, student services, researchers, and professional services staff. The consultation is open now, and Jisc says the pilot is expected to run across the 2026-27 academic year, split between a discovery stage and an optional practice stage. The practical takeaway is that this is not yet a finished framework, but it is already a sector signal that oversight claims will need more explicit definition.
What this means for institutions
First, universities should stop treating human review as self-evident. A person glancing at AI-generated feedback is not the same as a person being able to understand, challenge, edit, and reject it. Jisc's consultation suggests institutions will increasingly need to define who reviews outputs, what evidence they can see, what counts as a meaningful intervention, and how that judgement is recorded. That matters for assessment workflows, but it also matters wherever AI-generated or AI-summarised outputs can influence student-facing decisions.
Second, the proposal has a clear implication for student feedback evidence. This is an inference from Jisc's marking-and-feedback source, not a direct claim made by Jisc about survey analytics: if universities use AI to summarise module evaluations, service feedback, or survey comments, they will face the same oversight question. Who can inspect the source comments? Can a team trace a summary back to the underlying evidence? What happens if an output looks plausible but flattens disagreement or misses a safeguarding issue? Those are governance questions, not only technical questions, and they are harder to answer if teams rely on ad hoc generic LLM workflows.
Third, institutions should treat this as a near-term governance task rather than a future procurement task. Jisc's proposal is explicitly tool-agnostic and built around tools universities already have. That means Student Experience teams, PVCs, and quality leaders can start now: map where AI is already touching assessment, support, and evidence workflows; define where human oversight sits; and document how outputs will be checked, escalated, and retained. A practical starting point is a student comment analysis governance checklist, because the same disciplines that protect comment analysis also help universities scrutinise wider AI-supported evidence.
How student feedback analysis connects
Jisc's pilot is about AI in marking and feedback, not specifically about NSS or module evaluation analytics. Even so, the governance problem is familiar. When universities use AI to code, cluster, summarise, or prioritise open-text student comments, they still need to know what source material was in scope, what review step sat between raw data and reported conclusion, and how exceptions were handled. A method such as our NSS open-text analysis methodology is useful because it keeps coverage, traceability, and interpretation more explicit than a one-off summary pasted into a committee paper.
The broader lesson is simple: if AI-supported evidence is consequential, it should be reviewable, contestable, and attributable. Jisc's latest proposal pushes the sector in that direction. For universities using student feedback to support quality enhancement, assessment review, or institutional decision-making, that is the most important takeaway.
FAQ
Q: What should institutions do now if they are already using AI in student-facing or evidence workflows?
A: Start with an inventory. Identify where AI is already drafting feedback, summarising comments, supporting triage, or shaping decisions. Then document who reviews the output, what evidence they can inspect, when they are expected to challenge it, and how the final decision is recorded. If that process is still informal, use a checklist such as the student comment analysis governance checklist to turn it into a clearer operating model.
Q: What is the timeline and scope of Jisc's proposed pilot?
A: Jisc published the consultation article on 18 June 2026. The consultation is live now, and Jisc says the pilot is expected to run across the 2026-27 academic year. The proposal is open to Jisc member colleges and universities, and the intended participants include educators, assessment leads, quality and governance teams, student services, researchers, and professional services staff.
Q: What is the broader implication for student voice?
A: Human oversight is becoming something universities will need to evidence, not just assert. As AI reaches assessment, support, and comment analysis, student voice work will need clearer methods for showing how conclusions were reached, what was reviewed by people, and where staff judgement overruled automation.
References
[Jisc / National Centre for AI in Tertiary Education]: "What does “human in the loop” actually mean? Consulting on our next pilot idea" Published: 2026-06-18
[Jisc / National Centre for AI in Tertiary Education]: "Insights from the AI in Marking and Feedback Pilot" Published: 2026-05-20
[Jisc / National Centre for AI in Tertiary Education]: "University of Nottingham’s blind-study evaluation of AI in assessment design" Published: 2026-06-01
Request a walkthrough
Book a free Student Voice Analytics demo
See all-comment coverage, sector benchmarks, and reporting designed for OfS quality and NSS requirements.
- All-comment coverage with HE-tuned taxonomy and sentiment.
- Versioned outputs with TEF-ready reporting.
- Benchmarks and BI-ready exports for boards and Senate.
UK-hosted · No public LLM APIs · Same-day turnaround