Jisc's AI in assessment findings show student buy-in needs clearer communication
Updated Jun 06, 2026
assessment methodsartificial intelligenceAI in assessment is starting to look less like a tooling question and more like a communication and governance question. On 21 May 2026, Jisc published New findings highlight the benefits of a collaborative approach to AI in assessment, saying early lessons from its year-long pilot show that student buy-in depends on honest communication, visible human oversight, and clearly defined use cases. For Student Experience teams, PVCs, and quality professionals, that matters because universities will need sharper evidence on trust, clarity, and usefulness if AI-supported marking and feedback are going to move beyond small pilots.
What has changed in Jisc's AI in assessment findings
The 21 May update is a Jisc-wide summary of initial findings from its AI in marking and feedback pilots, rather than another single-use-case reflection. Jisc says the pilot involves 38 UK colleges and universities using education-specific tools from Graide, Keath, and TeacherMatic in assessment settings. The scope is therefore wider than one institution and wider than higher education alone, but the practical issues will be familiar to UK universities considering AI-supported feedback: where human judgement sits, how students are told about the pilot, and which parts of assessment are low enough risk to test first.
The headline lessons are specific. Jisc says keeping the human in the loop is harder in practice than it sounds, that open dialogue helps build student buy-in, and that AI should not be treated as a catch-all fix for marking and feedback pressures. It also repeats the point we covered in our earlier summary of Jisc's formative-first pilot findings: formative assessment remains the lowest-risk starting point because students can still use the feedback while learning is in progress.
"What struck me was how quickly conversations shifted from the AI tools to much deeper questions about assessment itself."
The newer operational detail sits in the additional outcomes. Jisc says pilot participants developed shared resources, including a student communication pack, because explaining AI use to staff and students was a common challenge. It also says feedback from participating institutions was used directly with product developers, and that clearer marking criteria and more explicit assessment frameworks improved the consistency of AI output. In short, the pilot is now producing implementation lessons about communication, rubric quality, and peer learning, not only lessons about tool capability.
What this means for institutions
The first implication is that an AI assessment pilot needs a communication plan as much as a technical plan. If students do not know whether AI is drafting comments, suggesting feedback, checking rubric alignment, or doing something closer to grading, trust will fall back on assumption rather than evidence. Jisc's update suggests universities should decide early what they will tell students, what questions they expect students to ask, and how staff will explain where academic judgement still sits.
The second implication is that peer learning and rubric design are becoming part of the risk-control model. Jisc's summary suggests institutions learned from each other, not only from vendors, and that AI output improved when marking expectations were defined clearly from the outset. That matters for quality teams because it shifts part of the implementation burden back onto assessment design. If criteria are vague, moderation is weak, or staff interpret rubrics inconsistently, AI-supported feedback will expose those weaknesses quickly rather than smoothing them away.
The third implication is that institutions should collect better student evidence before rollout widens. Jisc's findings sit neatly alongside the later OfS and Advance HE AI research project, which is also asking the sector for a stronger evidence base on how AI affects learning and assessment. Universities should therefore gather targeted student feedback on whether AI-supported comments feel clear, generic, fair, useful, or too detached from the module context. That gives leaders something stronger than anecdote when they decide whether a pilot should expand.
How student feedback analysis connects
This is where open-text analysis becomes more useful. Comments about AI in assessment rarely arrive under one neat heading. Students are more likely to talk about feedback quality, trust in staff judgement, clarity of permitted use, disclosure, fairness, and whether the comments actually help them improve. A workflow such as our student comment analysis governance checklist helps teams decide how those comments will be collected, reviewed, compared, and reported before the pilot reaches committee stage.
The next step is to analyse those comments consistently rather than reading them as one general reaction to AI. Our NSS open-text analysis methodology is a useful model because it keeps themes, source coverage, and action trails explicit. Student Voice Analytics can help institutions compare those patterns across pilots, modules, and survey routes, but the bigger point is methodological: once AI-supported feedback moves into live teaching, universities need evidence that can distinguish faster feedback from better feedback.
FAQ
Q: What should institutions do now if they are running or planning an AI in assessment pilot?
A: Write a short pilot brief that covers the use case, the human review step, the student communication approach, and the questions you will ask students afterwards. Then collect feedback early through module evaluations, pilot surveys, or representative channels, so trust and clarity issues surface while the pilot is still small enough to adjust.
Q: What is the timeline and scope of Jisc's latest AI in assessment update?
A: Jisc published the official update on 21 May 2026. It draws on a year-long pilot running from September 2025 to August 2026 and says the work involves 38 UK colleges and universities using Graide, Keath, and TeacherMatic in assessment settings. The scope is UK-wide tertiary education, but the operational implications are directly relevant to higher education providers.
Q: What is the broader implication for student voice?
A: Student voice on AI in assessment now needs to go beyond broad approval or disapproval. Universities need evidence on whether students understood the workflow, trusted the role of staff judgement, found the feedback actionable, and felt the process was fair enough to use at scale.
References
[Jisc]: "New findings highlight the benefits of a collaborative approach to AI in assessment" Published: 2026-05-21
[Jisc / National Centre for AI in Tertiary Education]: "AI in Assessment Pilot" Published: 2025-05-14
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