It is genuinely encouraging to see so many academics engaging with AI in education. Since the arrival of ChatGPT, the volume of publications, conference sessions, workshops, module redesigns and institutional conversations has grown rapidly. That energy matters. It shows that higher education is not standing still in the face of technological change; academics are experimenting, asking questions and sharing what they are learning.
At the same time, volume is not the same as value. As the field grows, a harder question emerges: how much of this work is helping us understand teaching and learning differently, and how much is repeating similar observations in different settings? The issue is that AI education scholarship can become fragmented when local pilots, student surveys, assessment redesigns and institutional guidance are not connected to a wider agenda. Practices may be duplicated rather than developed, and innovation may remain in isolated pockets rather than contributing to cumulative knowledge.
This concern sits within a wider debate about AI hype and scholarly overload. GenAI is not only changing education; it is also changing how academic work is produced and circulated. Recent evidence suggests that AI may be amplifying a “more rather than better” research culture, with one study reporting a 42% rise in submissions after ChatGPT alongside declining writing quality linked largely to AI-generated writing. For AI education, this raises an important challenge: how do we move from rapid activity to scholarship that has lasting educational value?
There is also a methodological point. Much current AI education work is understandably exploratory, often relying on surveys of staff or student perceptions. These studies are useful because they show how people are experiencing AI. But they cannot be the whole evidence base. If we want to know whether AI is improving learning, judgement, feedback or assessment, we also need richer forms of evidence: analysis of student work, assessment outcomes, focus groups, observations, learning analytics and design-based studies that follow change over time.
This is where the Scholarship of Teaching and Learning (SoTL) becomes important. Many academics already care deeply about teaching and regularly try new approaches in their classrooms. SoTL builds on that instinct, but takes it a step further by turning teaching questions into evidence-informed contributions that others can understand, use and build on. Individually, it helps academics sharpen the value of their work. Institutionally, it helps universities learn from local inquiry rather than relying only on policy statements or disconnected examples of practice.
We argue that the next stage of AI education research should be less tool-led and more question-led, moving beyond reactions to GenAI towards clearer educational purposes, stronger evidence and more connected inquiry across contexts. Our six future research directions can therefore be read as practical entry points for academics who want their work on AI education to contribute more meaningfully to the field.
1. Purpose
What should intelligent technologies help students and educators achieve?
Future research should focus on the educational purposes of intelligent technologies rather than the technologies themselves. The key question is how AI and related systems can support learning, creativity, critical thinking, participation and human development.
2. Context
How do intelligent technologies affect different educational settings?
Future studies should explore how intelligent technologies shape learning across schools, universities, professional education and lifelong learning. Research should compare how educational needs, challenges and opportunities differ across contexts.
3. Technology
What kinds of intelligent technologies are being studied?
Research should move beyond focusing only on GenAI and examine a wider range of intelligent technologies, such as recommender systems, feedback tools, analytics systems and conversational agents. Studies should also explore the assumptions and impacts embedded within these technologies.
4. Pedagogy and Assessment
Under what conditions do intelligent technologies support learning and assessment?
Future research should investigate how intelligent technologies influence teaching, learning, feedback, assessment, integrity, inclusion and learner agency. Comparative studies of authentic assessment and AI-enabled learning approaches would be especially valuable.
5. Capacity and Governance
What support, leadership and governance are needed?
Research should examine how institutions build staff capability, leadership, policy and governance structures for responsible and sustainable AI adoption. This includes professional development, organisational readiness and ethical oversight.
6. Method
What research approaches are needed?
Future studies should include more longitudinal, comparative and cross-institutional research. Greater use of collaborative and participatory research involving educators, students, employers and policymakers would help build a stronger and more inclusive evidence base.
Taken together, these six directions are not a checklist. They are quality filters for a crowded field. They ask whether our work has a clear purpose, responds to a real context, examines a specific technology, improves pedagogy or assessment, builds institutional support and rests on appropriate evidence.
The key message is that moving beyond AI hype does not mean producing more scholarship for its own sake.
It means building scholarship that lasts.
In a crowded field, the most valuable contribution is not always another output, but a clearer question, stronger evidence and a more useful account of what others can learn.
Meaningful SoTL enables universities to build sound governance and provide strategic direction for integrating AI in a purposeful and impactful way.
Dr Marios Kremantzis is Senior Lecturer in Business Analytics at the University of Bristol Business School, a Senior Fellow of Advance HE and CMBE. A decision scientist and award-winning educator, his work focuses on AI-enhanced business education, decision analytics, performance measurement and responsible assessment, alongside academic leadership in postgraduate education and pedagogic scholarship.
Professor Xue Zhou is Dean of AI at the University of Leicester’s College of Business and a Professor in AI in Business Education. A Principal Fellow of Advance HE and award-winning educator, her work focuses on AI literacy, digital pedagogy and ethical AI use. She leads funded projects and international initiatives advancing AI innovation in education and practice.
Dr Aniekan Essien is Senior Lecturer and Programme Director for Business Analytics at the University of Bristol Business School. An AI and machine learning researcher and data scientist, his work applies analytics, deep learning and technologies to supply chains, operations, innovation management and business improvement, advancing practical impact through technology.