Dr Anthony Beh Kuan Jun, Dr Jessica Price, Dr Tissa Chandesa and Dr Wei Keong Too from the University of Nottingham Malaysia, discuss how universities can lead by learning from practice as it emerges.

“My institution has not told me what to do.”  

This frustration is understandable. Many educators feel they have been handed responsibility for responding to GenAI without the time, resources or support needed to do so. For colleagues already balancing teaching, research, administration and student support, GenAI can feel less like an opportunity but another major change to absorb into an already crowded workload.  

But waiting for complete instructions may also be the wrong place to begin. GenAI is changing too quickly, and affecting disciplines differently, for universities to prescribe a single response. To understand why institutional guidance alone cannot resolve the disruption, we first need to consider what GenAI changes about learning itself.  

The appearance of understanding  

The deeper challenge is that GenAI systems change the relationship between the learner and knowledge.  

Before GenAI, students would usually encounter disciplinary knowledge through a process guided by an educator. Lecturers carefully scaffold concepts, provide frameworks, model academic judgement and help students learn how to organise and evaluate information.  

A naïve use of GenAI can make learning and understanding appear effortless. A student can enter a question and receive a fluent, confident and apparently competent response within seconds. This results in a condition the literature has coined, epistemia (producing the feeling of knowing without the labour of judgement).  

The concern is not simply that these systems may at times produce inaccurate information. It is that students may get a sense of achievement or competence when using the AI to produce something of value, without the necessary intellectual effort that typically accompanies this kind of work. This is often signalled by the failure to explain, defend or independently reproduce the thinking behind it without the help of AI later.  

By contrast, a human expert could provide students with a similarly competent answer, but the relationship is fundamentally different. In traditional pedagogical models, students can observe and gradually internalise deeper forms of judgement and reasoning through iterative exchanges with an educator. This process is not inherently replicable with GenAI in its unguided state.   

The GenAI disruption is neither uniform nor isolated  

The effects of generative AI are not the same across every discipline, programme or form of assessment. In some areas, AI may primarily challenge established approaches to writing and assessment. In others, it may change how students code, design, analyse data, generate ideas, communicate with people or prepare for professional practice. Some disciplines may be able to integrate these tools relatively quickly. Others face greater challenges because their assessments are closely connected to professional accreditation, disciplinary conventions or particular forms of human judgement.  

The disruption is therefore not uniform. It is also not isolated.  

Universities often address AI through separate concerns about assessment, policy, classroom practice and employability. Yet a solution in one area may create problems elsewhere. Measures designed to restrict inappropriate use may also limit opportunities for students to develop capabilities expected in the workplace. Similarly, institutions may promote AI literacy without considering whether their definition reflects how employers expect graduates to use these systems.  

Universities therefore risk preparing students for the wrong environment if these conversations remain disconnected. This is why a response narrowly designed to address assessment or policy is likely to be short-lived. GenAI is changing too quickly and its effects reach too far across the educational system for one isolated solution to remain effective.  

The importance of inviting different perspectives  

These concerns helped motivate our conference, Understanding Generative AI: What Educators Need to Know to Redesign Learning, held in Kuala Lumpur, Malaysia in July 2026.  

 We wanted to move beyond isolated discussions of assessment, policy or classroom practice and examine how these issues connect. The conference  brought together 48 participants, including colleagues from external institutions and industry, to explore GenAI through perspectives spanning institutional leadership, teaching, assessment, educational technology and industry.  

As the event progressed, it became clear that we were converging on similar themes, albeit from vastly different perspectives and disciplinary backgrounds. Institutional and policy discussions highlighted gaps in readiness and common blind spots, where the most visible challenges attracted attention while less obvious but equally important issues were overlooked. Classroom examples showed how meaningful change often begins with educators testing new approaches in practice within their discipline. Industry perspectives provided a broader view of how quickly expectations of graduate capability are shifting.  

The conference concluded with a workshop that asked participants to bring these perspectives together and consider what they could apply within their own settings. Our post-event feedback suggested that many were already considering changes to their teaching and assessment practices and showed stronger confidence across several domains of AI readiness.   

One lesson from our post-event review was the need for a stronger student voice. Students are directly affected by AI policy while also preparing to enter workplaces already being reshaped by these systems. They therefore need a seat at the table when institutions decide how to respond.   

A whole-institution response does not mean one size fits all  

A whole-institution response should not require every discipline to use GenAI in the same way. Instead, leadership should establish broad principles, clarify the institution’s position and provide time, resources and permission to experiment. Disciplines and individual educators can then determine how those principles apply within their own forms of teaching, assessment and professional practice.  

Much of this change will need to develop through trial and error. An educator might test a new activity, redesign part of an assessment or ask students to make their reasoning and AI use more visible. What works and what does not should then be shared rather than remaining within one classroom. Institutions can connect colleagues facing similar challenges, evaluate emerging practices and use them to refine wider guidance.  

Over time, these small experiments can move beyond isolated examples. Shared practices become common expectations, and common expectations gradually become institutional culture. This allows change to grow from practice while remaining supported by a clearer institutional direction.  

This brings us back to the opening frustration. An institution may not be able to tell every educator exactly what to do, but it can make it possible for them to begin. The practical starting point is not to wait for the perfect policy to present itself or redesign an entire programme. Start with one change: one activity, one assessment or one conversation with students.  

How has GenAI already changed learning in your classroom and what is one change your institution could support today?