AI chatbots can speed up student support, but universities still need to track trust and usefulness
Updated Jul 09, 2026
Students do not judge support systems by whether they are innovative. They judge them by whether they solve a problem quickly, accurately, and without sending them in circles. At Student Voice AI, we see the same test in open comments about digital services and academic support. That is why Marios Kremantzis, Anthi Chondrogianni and Aniekan Essien's Journal of Further and Higher Education paper, "Evaluating the impact of AI chatbots on student support and engagement in UK higher education", matters. Like recent evidence on how students judge AI-using teachers by care and responsibility, it suggests that AI services in higher education are judged relationally, not only technically.
Context and research question
Student support teams are under pressure from rising query volumes, fragmented digital systems, and student expectations of rapid answers. Chatbots look attractive in that environment because they promise scale and constant availability. The harder question for UK universities is whether they actually make support easier to use, or whether they simply add another layer that students must navigate before reaching the right person.
This study focuses on Business and Economics cohorts in UK higher education, across both undergraduate and postgraduate levels. The authors deployed a rule-based, decision-tree chatbot designed to provide general and unit-specific information, then evaluated it using quantitative engagement metrics and qualitative student feedback. That design makes the paper useful for Student Experience teams because it looks beyond implementation and asks what students actually found helpful.
Key findings
The strongest operational finding is that the chatbot resolved a meaningful share of routine queries. The abstract reports an estimated 59% success rate in resolving student questions, covering both administrative and academic enquiries. That matters because many support requests are not high-complexity cases. They are questions about where to find information, how a process works, or which route to use next. If a chatbot can absorb that first layer reliably, staff time can be redirected towards the cases that genuinely need judgement or pastoral care.
Students valued immediacy very strongly. The abstract reports that 91% of respondents successfully located relevant information, which suggests the tool reduced search friction as much as it answered questions. For UK teams, that is an important distinction. A support system can fail even when the information technically exists, if students cannot find it quickly inside module pages, handbooks, and service portals.
"100% of respondents valued the expedited query resolution"
That short quote captures the paper's clearest message. Students appear to have valued speed in its own right, not as a cosmetic feature but as a practical reduction in effort. When routine uncertainty is resolved quickly, the surrounding student experience can feel more navigable.
The engagement result is promising, but it needs careful interpretation. The abstract states that 74% reported increased engagement with their courses after using the chatbot. That is notable because universities often separate support from engagement, treating one as service provision and the other as a teaching matter. This paper suggests the boundary is thinner than that. When students can get answers faster, they may stay closer to the course. At the same time, the figure is self-reported, and the abstract does not specify the respondent count behind the feedback percentages, so UK universities should read it as strong local evidence rather than a sector benchmark.
The broader takeaway is that chatbots work best as a complement, not a replacement. The abstract frames the chatbot as something that can extend traditional educational resources through prompt, accessible support. That is a useful corrective to more inflated AI narratives. The paper does not suggest that human support is obsolete. It suggests that students benefit when simple questions are answered quickly and the surrounding support environment becomes easier to navigate.
Practical implications
For UK higher education teams, the first implication is to treat chatbots as service design, not just AI procurement. Decide in advance which query types should be answered instantly, which should trigger handoff to staff, and which should never be automated. Questions about unit information, deadlines, and signposting are much more plausible candidates than sensitive wellbeing issues or complex academic judgement. The benefit is faster access without blurring where human support begins.
Second, universities should evaluate chatbot rollouts with student feedback, not usage data alone. A dashboard can show query volume and resolution rates, but it cannot tell you whether the answer felt clear, trustworthy, or reassuring. Build short follow-up prompts and open-text questions into the rollout, much like York's digital module evaluation system shows the value of faster feedback loops and visible response processes. That gives Student Experience teams earlier evidence on what is helping and what is still sending students back into staff inboxes.
Third, institutions should segment the evidence by cohort and query type. This study covers undergraduate and postgraduate Business and Economics students, which is useful because support pressures are rarely uniform. An international postgraduate student, a commuter undergraduate, and a final-year student approaching deadlines may all use the same chatbot for different reasons. If universities want actionable evidence, they should compare which groups use the tool, which questions remain unresolved, and which comments point to reassurance, confusion, or unmet need. The benefit is more targeted support improvement rather than one generic verdict on the tool.
Finally, governance still matters, even when the system is rule-based rather than generative. Teams need to know what data is logged, who reviews unresolved cases, how sensitive queries are handled, and how updates to the chatbot's knowledge base are approved. If institutions want chatbot feedback to count as usable evidence rather than anecdote, a student comment analysis governance checklist is a sensible starting point. The benefit is not just lower risk. It is clearer accountability when students say a support tool helped, confused, or excluded them.
FAQ
Q: How should a university pilot a student-support chatbot without weakening human support?
A: Start with a narrow use case, such as module information, deadline signposting, or standard service queries, and publish clear boundaries on what the chatbot can and cannot do. Then collect one short open-text prompt after interactions, asking what felt helpful and what still required staff input. If the chatbot sits inside a wider AI strategy, Jisc's June HE AI meetup on student feedback guidance is a useful reminder to ask more precise questions about clarity, consistency, and trust.
Q: What should teams keep in mind before generalising from this study?
A: The study is UK-based, which increases its practical relevance, but it is still one implementation in Business and Economics across undergraduate and postgraduate cohorts. It evaluates a rule-based, decision-tree chatbot rather than a generative AI assistant, and the abstract reports headline percentages without giving the full respondent count. Universities should therefore treat it as a strong pilot signal, not proof that every chatbot will improve engagement in the same way.
Q: What does this change about student voice practice more broadly?
A: It widens the scope of what counts as student voice evidence. If support increasingly happens through digital tools, then student comments about those tools become part of the institutional evidence base, not a side issue for IT. Universities should ask not only whether a chatbot was used, but whether it reduced friction, improved confidence, and helped students stay connected to their course.
References
[Paper Source]: Marios Kremantzis, Anthi Chondrogianni and Aniekan Essien "Evaluating the impact of AI chatbots on student support and engagement in UK higher education" DOI: 10.1080/0309877x.2026.2693084
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