Student voice on AI is more conflicted than university policy assumes
Updated Jul 25, 2026
Universities often talk about student views on Generative AI as if students fall into neat camps: enthusiastic adopters, anxious resisters, or policy rule-breakers. In reality, the same student can move between all three positions depending on the task, the stakes, and the guidance they receive. That is why Tim Fawns, Margaret Bearman, Thomas Corbin, Michael Henderson, Jan McLean, Kelly E. Matthews, Glenys Oberg, Yifei Liang and Jack Walton's Higher Education paper, "Illuminating complex student realities of artificial intelligence through an entangled pedagogy framework", matters. For UK teams reviewing AI-related module comments, pilot feedback, and wider student experience data, it adds useful depth to recent Advance HE evidence on student experiences of GenAI in UK universities.
Context and research question
Many institutional AI policies still assume that the main task is to separate acceptable from unacceptable use. That is necessary, but it is not enough. Students are making judgements about AI in relation to learning, identity, fairness, workload, confidence, and future employability all at once. If universities only ask whether students approve of AI, they risk flattening those tensions into a single headline view.
This paper tackles that problem through a qualitative design. Drawing on Fawns' entangled pedagogy framework, the authors focus on three participants from a broader set of student focus groups and examine how those students made sense of GenAI in relation to their studies, identities, and everyday practices. That matters for UK higher education because it shifts the question from "Are students for or against AI?" to "What conditions make AI feel useful, risky, unfair, or misaligned with learning?"
Key findings
The paper's clearest contribution is to show that students' views on GenAI are not fixed positions. They shift across contexts, purposes, and pressures. A student may see AI as helpful in one moment and troubling in another, which means institutions should be cautious about treating a single survey item or one approval score as a settled judgement. For Student Experience teams, that is a practical warning against over-reading blunt AI sentiment measures.
Students' responses were entangled with what they thought learning was for. The abstract shows that their views were shaped by conceptions of learning, moral positions, and contextual considerations, not only by the tool itself. In practice, that means comments about AI are often also comments about effort, authorship, independence, and the kind of learner a student is trying to be. That sits closely with recent evidence that students judge AI-using teachers by care and responsibility, not only by whether the technology appears efficient.
Institutional governance was part of the student experience, not just a background rule set. The abstract points to students' relationships with AI being shaped in relation to institutions, educators, peers, and perceived future trajectories. That matters because students do not encounter AI as a neutral study aid. They encounter it through local rules, uneven staff signals, and assumptions about what counts as legitimate academic work. Where those signals are inconsistent, student comments are likely to show tension long before a policy review catches up.
"appropriate institutional responses must involve ongoing dialogue with students"
The paper also highlights contradiction as useful evidence rather than noise. By examining just three participants in depth, the authors were able to surface tensions that larger surveys can miss: students can value AI's usefulness while distrusting its effects, or welcome flexibility while worrying about what it does to their voice or judgement. For UK teams, that is a reminder that disagreement and ambivalence in open comments are not a reporting problem. They are often the most informative part of the evidence.
The broader message is that universities need to listen for complexity, not force premature consensus. This is not a call for indecision. It is a call to build AI policy and teaching guidance around how students actually navigate competing values in practice. That gives institutions a better chance of responding to real concerns, rather than to an oversimplified version of student opinion.
Practical implications
For UK universities, the first implication is to stop asking generic questions about AI use. Instead of only asking whether students support or oppose GenAI, ask where it felt useful, where it felt risky, and where it seemed to pull against the purposes of learning. A better prompt is often contextual, for example asking how AI affected drafting, feedback, confidence, or the sense of doing one's own thinking. That produces evidence teams can actually act on.
Second, institutions should treat AI-related student comments as governance evidence as well as experience evidence. If students describe uncertainty, mixed staff messages, or changing rules between modules, that is not simply confusion at the edge of implementation. It is part of the lived quality of the course. This aligns with QAA's July 2026 warning that inconsistent AI practice has become a student experience risk. The practical gain is earlier visibility of where policy drift is undermining trust.
Third, universities should design student voice processes that can capture contradiction without forcing students into a single position. Short pulse surveys, structured open-text prompts, and focused follow-up conversations can all help. The aim is not to collect more AI opinion in the abstract, but to understand where guidance, assessment design, and teaching practice are creating friction. That is where Student Voice Analytics fits naturally: it helps teams group AI-related comments into themes such as trust, fairness, self-expression, authorship, and policy clarity, so local debates rely less on anecdote and more on pattern-level evidence.
Finally, institutions should document how AI-related comments will be reviewed, challenged, and acted on. If AI now cuts across teaching, assessment, and student support, then the evidence derived from student comments needs a clear audit trail too. A practical starting point is a student comment analysis governance checklist, because the same disciplines that improve comment analysis also improve confidence in how AI-era student evidence is handled.
FAQ
Q: How should a university collect more useful feedback on student experiences of GenAI?
A: Use context-specific questions rather than broad approval items. Ask where students found AI helpful, where they felt uncertain, and which rules or teaching practices made responsible use easier or harder. One good open-text prompt is to ask what made AI feel supportive in one context and risky in another. That gives teams a better basis for revising module guidance, staff development, and assessment communication.
Q: What are the methodological limits of this paper?
A: This is a qualitative paper that focuses closely on three participants from a broader focus-group project. Its strength is depth, not representativeness. Universities should not treat it as a benchmark for how all students think about AI. They should use it as an interpretive framework for reading their own local student comments, surveys, and follow-up conversations with more nuance.
Q: What does this change about student voice practice more broadly?
A: It suggests that AI-related student voice should not be read as a simple vote for or against technology. Comments about AI often carry signals about trust, effort, fairness, identity, and policy clarity at the same time. That is why institutions need sharper student voice design on AI, much like the challenge outlined in Jisc's June 2026 discussion on collecting clearer student feedback about AI guidance. The benefit is clearer evidence on what needs to change and why.
References
[Paper Source]: Tim Fawns, Margaret Bearman, Thomas Corbin, Michael Henderson, Jan McLean, Kelly E. Matthews, Glenys Oberg, Yifei Liang and Jack Walton "Illuminating complex student realities of artificial intelligence through an entangled pedagogy framework" DOI: 10.1007/s10734-026-01730-1
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