Jisc AI guidance sets boundaries for student comment analysis

Updated Nov 06, 2026

Jisc's AI guidance published on 23 October 2026 explicitly addresses analysing student survey comments within learning analytics. For university survey leads and Student Experience teams, it raises a practical question: does the proposed use of a comment fit the purpose for which students supplied it?

What the Jisc AI guidance adds

The October article accompanies Jisc's code of practice for learning analytics, whose page records an update on 9 September 2026. The code addresses UK educational institutions and applies regardless of the technology used. The October companion explains how those principles apply to AI.

The companion connects analysis of written student content with limits on collection and use. Its data-minimisation section states:

"student content should only be analysed when it is provided for that purpose"

It also recommends staff checking of AI interpretations and a meaningful route for students to question or correct information used.

What this means for institutions

The code recommends consulting student representatives during design, rollout and monitoring, explaining AI use, and validating outputs through qualified staff. It also warns about re-identification when datasets are combined. These points make survey governance relevant before comments enter an analytics system.

Our recommendation is to review a proposed use with the survey owner, analytics team and student representatives. Write down the collection purpose, intended decision, permitted audience and review process. Ask students whether the explanation makes sense from their perspective. Our guide to student voice and shared decisions offers a starting point for deciding how that discussion should influence the design.

For example, separate a proposal to review recurring difficulties with a support service from a proposal to attach comments to individual attendance records. Document the purpose and safeguards for each, including whether the original invitation supports that use. Do not attempt to identify the author of an anonymous response to make a linkage possible. Agree the boundary before selecting or configuring a tool.

How student feedback analysis connects

For an aggregate service review, we suggest keeping the survey question and collection context alongside each theme. Ask reviewers to check whether a summary preserves mixed experiences, such as helpful advice followed by an unresolved referral. Record corrections and unresolved interpretations before the report reaches a committee. Our open-text analysis methodology explains how to retain context and trace a theme back to the comments supporting it.

Make the review step concrete: name who can inspect the supporting comments, reject an interpretation and approve the report. Keep a record of the version reviewed and the changes made. The student comment governance checklist provides questions about access, retention, validation and reporting. Use the analysing comments pathway to connect those checks with the wider reporting process.

FAQ

Q: What should university teams do now?

A: Review an existing or proposed comment-analysis workflow. Check its stated purpose, data access, human review and route for handling disputed interpretations. Test the explanation with students before extending the workflow to another use.

Q: Who should own the review?

A: Name a person who understands the survey context and has authority to question the analysis. Give them access to the supporting evidence and a clear route to resolve concerns before the findings are shared.

Q: What does this mean for student voice?

A: The code gives students a role in shaping analytics and questioning AI interpretations. Our recommendation is to include that opportunity in feedback-system design, then review with students whether the route is understandable and usable.

References

[Jisc]: "The ethical application of AI in learning analytics" Published: 2026-10-23

[Jisc]: "Code of practice for learning analytics" Published: 2015-06-04; updated: 2026-09-09