At this year's UCISA conference, one session captured a challenge that many universities are quietly grappling with as AI moves from experimentation to everyday reality. Presented by Wan Ting Wu from the University of Glasgow, the research explored how students are navigating Generative AI in higher education, what concerns them, how they are using it, and what support they need from their institutions. While much of the sector's AI conversation over the past two years has focused on governance, academic integrity and risk, the findings suggest that universities may now be confronting a different challenge altogether. Students are no longer asking whether they should be using AI. Instead, they are trying to understand how to use it appropriately, ethically and confidently within the context of their learning.
This distinction is important because it signals a shift in where the sector's attention needs to be focused. The first phase of the AI debate was understandably dominated by questions about acceptable use, plagiarism, accuracy, privacy and academic standards. Universities needed to establish policies, define boundaries and create governance frameworks. However, as AI tools become increasingly embedded in student life, the challenge is no longer simply determining whether AI belongs within higher education. For many students, that question has already been answered. The more pressing issue is whether universities can create a sufficiently consistent environment to help students use these tools effectively, responsibly and with confidence.
One of the most interesting findings from the research was the role peer networks play in driving AI adoption. Many students reported that they only began using AI after observing classmates and friends using it successfully. In other words, AI behaviours are spreading informally through student communities rather than primarily through institutional channels. This trend is particularly significant because it highlights how quickly students are adapting to new technologies, often independently of formal guidance or curriculum design.
For generations, universities have been the primary source of guidance on how knowledge is accessed, evaluated and applied. AI is beginning to change that dynamic. Students are increasingly arriving with experience of AI tools, established habits and pre-existing expectations about how these technologies can support learning. Rather than introducing students to AI, institutions are increasingly encountering students who have already begun experimenting with it themselves. This does not diminish the university's role, but it does change it. Instead of acting solely as gatekeepers, universities must now become enablers, helping students understand how to use AI effectively while developing the critical thinking and academic judgement needed to apply it responsibly.
The implication is that universities no longer have the luxury of treating AI as a future issue. Students are already engaging with these technologies today, often in ways that are influenced more by peers than by institutional guidance. This creates a growing need for universities to establish a clear and consistent position that helps students navigate AI with confidence rather than uncertainty.
Perhaps the most revealing aspect of the research was the relationship between institutional policy and day-to-day student experience. Students reported looking to university policies for guidance, but they also rely heavily on the attitudes and behaviours of lecturers when interpreting what is and is not acceptable. When institutional policies and academic practice are aligned, students gain confidence. When they are not, confusion quickly emerges.
This points to what is arguably the most significant institutional challenge emerging from AI adoption. Most universities have already invested considerable effort in developing governance frameworks, creating guidance documents and establishing principles for responsible AI use. These activities remain essential, but the research suggests that policy alone is not enough. A policy can define acceptable use, but it rarely answers the practical questions students face every day. How should AI be used when researching an assignment? What level of support is appropriate when drafting coursework? Where does productivity enhancement end and academic misconduct begin? These are the questions that shape behaviour in practice, and they cannot be resolved through governance alone.
The challenge facing many institutions is therefore not one of policy creation but policy operationalisation. Universities may have a clearly defined position on AI at an institutional level, but unless that position is translated consistently into teaching practice, assessment design and student support, confidence will remain elusive. What the Glasgow research ultimately highlights is a gap between institutional intent, academic implementation and student experience. The task now is finding ways to close that gap.
The research also revealed something that is often overlooked in public discussions about AI. Students are not blindly embracing the technology. They recognise many of the risks that concern educators and policymakers. Participants highlighted issues including academic integrity, hallucinations, over-reliance, privacy, reduced critical thinking and the potential for generic or formulaic outputs. Some even reported that validating AI-generated information can create additional work rather than saving time.
What makes this particularly interesting is that students are not asking universities to remove AI from the learning experience. Instead, they are asking for greater clarity around how to use it well. The research showed that students consistently want clear boundaries, practical examples, discipline-specific guidance and consistent messaging from academic staff. Policies remain important, but students appear to be looking for something more tangible: examples of what good AI use actually looks like within the context of their studies.
This reflects a broader shift in the sector's AI journey. The first phase was necessarily focused on governance and risk management. The next phase will be characterised by enablement. Universities will increasingly need to build confidence among both staff and students, not simply through policies but through training, practical examples and embedded support. The institutions that make the greatest progress are likely to be those that move beyond abstract guidance and provide people with the tools and confidence to apply AI appropriately within their own disciplines, roles and responsibilities.
The concept that emerged most strongly from the session was consistency. Every challenge identified by the research can ultimately be traced back to differences in understanding, confidence and interpretation. Students encounter varying guidance from different lecturers. Staff possess different levels of confidence in their own understanding of AI. Disciplines develop different attitudes towards acceptable use. The result is an inconsistent experience that can leave students uncertain about how they should engage with the technology.
This matters because inconsistency creates friction at every level. Students become hesitant to use tools that might otherwise enhance learning. Academics become uncertain about how to advise learners. Institutions struggle to realise the potential benefits of AI because adoption occurs unevenly across departments and faculties. While AI itself may be a technology challenge, successful adoption increasingly looks like an organisational challenge centred on alignment, communication and change management.
The importance of consistency extends well beyond teaching and learning. Universities are already exploring the role of AI within student recruitment, admissions, support services, research and operational processes. As these initiatives scale, institutions will need to ensure that students, academics and professional services teams all share a common understanding of where AI adds value, where the risks lie and what good practice looks like. The universities that succeed will not necessarily be those with the most sophisticated technology stack or the most comprehensive governance framework. They will be those that can translate strategic intent into everyday practice.
The Glasgow research provides an important reminder that AI readiness is not primarily a technology issue. It is an organisational issue. Students want confidence. Academic staff want confidence. Institutional leaders want confidence that AI is being used responsibly and effectively. Building that confidence requires more than policy documents. It requires alignment between strategy, guidance, teaching practice and student experience.
Perhaps the most powerful observation from the session was captured in the discussion that followed. Students already receive plenty of messages about what they should not do. What they increasingly want is guidance on what they should do instead. That insight feels particularly relevant as higher education enters the next phase of AI adoption. The sector has spent much of the last two years focused on control, governance and risk mitigation. The years ahead are likely to be defined by confidence, capability and operational consistency.
Universities do not have an AI adoption problem. Students are already adopting AI. The challenge now is ensuring that adoption happens in ways that are responsible, valuable and consistent across the institution. Those that succeed will be the organisations that move beyond policy alone and create an environment where staff and students understand not just the rules of AI, but how to apply it confidently to enhance learning, teaching and the wider university experience.