This article was authored by Dr Linda Marron PFHEA, with assistance from AI.
Artificial intelligence is changing far more than assessment design. It is prompting higher education to rethink what we value in learning, how we recognise achievement and how we prepare students for an increasingly AI-enabled world. Drawing on discussions from the AI & Authentic Assessment in a Digital Era conference, this article reflects on three shifts that are beginning to reshape assessment across the sector.
The conversation is changing
For much of the past three years, conversations about artificial intelligence in higher education have been dominated by a single concern: how do we preserve academic integrity?
The arrival of generative AI prompted understandable questions about plagiarism, authorship and assessment security. Institutions responded quickly, introducing policies, guidance and detection tools while educators adapted assessments to an educational landscape that seemed to change almost weekly.
Yet the conversation is beginning to mature.
Across higher education, educators are asking different questions. Rather than focusing solely on whether students use AI, attention is turning to how they use it, why they use it and, perhaps most importantly, what learning remains visible when AI becomes part of the educational process.
These questions formed the backdrop to AI & Authentic Assessment in a Digital Era, an online conference hosted by the Oxford International Digital Institute (OIDI). Bringing together academics, educational developers, learning technologists and institutional leaders from across the UK and internationally, the event explored emerging research, institutional practice and practical approaches to assessment and AI literacy in higher education. More importantly, it demonstrated that while institutions may be approaching AI from different starting points, many are arriving at remarkably similar conclusions.
The conference formed part of the AI Literacy & Authentic Assessment in Higher Education Collaborative Enhancement Project (CEP), funded by the Quality Assurance Agency for Higher Education (QAA) and led by the OIDI.
The project brings together a network of universities and institutions to explore how AI can be used in teaching, learning and assessment for both staff and students, with a particular focus on building confidence, capability and critical understanding.
The conference provided an opportunity to showcase emerging work from across the sector and contribute to ongoing conversations about the future of assessment in higher education.
What struck me throughout the day was not that every presenter shared the same perspective. Quite the opposite. Speakers represented different disciplines, institutions and educational contexts, each bringing their own experiences of integrating AI into teaching, learning and assessment. Yet despite these different starting points, several common themes emerged.
Rather than presenting a single solution, the conference became a collaborative exploration of how higher education is beginning to rethink assessment, learning and graduate capability in response to AI.
Across keynote presentations, workshops, demonstrations, case studies and panel discussions, one message became increasingly clear. The sector is no longer simply responding to AI. It is beginning to redefine what effective teaching, meaningful assessment and graduate capability look like in an AI-enabled world.
“The discussion is no longer centred on whether AI belongs in higher education. It is increasingly about how institutions can use it to strengthen learning rather than simply protect assessment.”
We are moving from AI adoption to AI maturity
AI strategy is becoming educational strategy
When generative AI first entered mainstream education, institutional responses were necessarily reactive. Policies were drafted, guidance produced and discussions centred on risk, compliance and academic misconduct. Institutions had little choice. The technology had arrived rapidly, and educators were understandably trying to understand both its potential and its risks.
That position is beginning to change.
Opening the conference, Kenrick Cabrera, Chief Technology Officer at Oxford International, argued that successful AI adoption is no longer simply a technology question. Instead, institutions need to think strategically about governance, organisational readiness, staff capability and measurable educational impact. Oxford International’s own approach, centred around technology enablement, governance and compliance, business alignment, training and skills development, illustrated how AI is increasingly becoming an institutional capability rather than a standalone technology initiative.
This perspective resonated because it reflected a broader shift taking place across higher education. AI implementation is no longer judged by the number of tools available to staff or students. Instead, institutions are beginning to consider whether they have created coherent approaches that balance innovation with educational purpose, ethical responsibility and organisational confidence.
The conversation has matured from “Which tools should we use?” to “What kind of institution do we want to become in an AI-enabled landscape?“
Innovation is becoming more intentional
This institutional perspective was reinforced by Stuart Feltham and Peter Beaumont from Edge Hill University, who shared their experience of introducing AI functionality within Blackboard Ultra. Rather than enabling every available feature as soon as it became available, Edge Hill adopted a carefully considered approach, evaluating each tool against pedagogical need, institutional priorities and student experience. Their presentation demonstrated that successful innovation is often characterised not by rapid implementation, but by thoughtful decision-making.
This reflects a wider change occurring across the sector.
The early stages of AI adoption were understandably characterised by experimentation. Institutions explored possibilities, trialled platforms and responded quickly to emerging technologies. Increasingly, however, the focus is shifting towards intentional adoption, asking not whether AI can be used, but whether it should be used, where it adds genuine educational value and how it aligns with institutional strategy.
Several speakers returned to this idea throughout the day. Whether discussing governance, learning technologies or curriculum design, the emphasis was consistently on purposeful implementation rather than technological novelty. AI was presented not as an end in itself, but as one component within a broader educational strategy.
Embedding AI where learning happens
This philosophy was equally evident in the work presented by Helen Wilson and Tom Ledbury from Bangor University International College. Rather than teaching AI as a separate topic, they described how it had been integrated into the redesign of a core IT and Communications module alongside digital literacy, professional communication and academic integrity. Students were encouraged to use AI appropriately within authentic assessment tasks while simultaneously developing the professional behaviours and digital capabilities expected beyond university.
This distinction is significant because it reframes AI literacy as something developed through authentic educational experiences rather than delivered through standalone training. Students build confidence not simply by learning about AI, but by using it responsibly within realistic learning contexts that require professional judgement, communication and critical reflection.
Taken together, these presentations suggest that institutional maturity is no longer defined by whether AI has been adopted. Instead, maturity is demonstrated by how thoughtfully AI is integrated into teaching, learning and assessment, supported by governance, staff development and a clear educational rationale.
Perhaps most importantly, this first section of the conference reminded us that successful AI adoption is not owned by technology teams alone. It requires collaboration between senior leaders, academics, educational developers, learning technologists and professional services. AI is increasingly becoming an institutional capability, and that capability depends as much on people and culture as it does on technology itself.
“Successful AI adoption is no longer about selecting the right tools. It is about creating the conditions in which staff and students can engage with AI thoughtfully, critically and ethically.”
We are beginning to assess thinking, not simply outputs
Moving beyond AI detection
If one idea connected more presentations than any other throughout the conference, it was this: assessment is increasingly moving beyond measuring finished products towards understanding the thinking that produced them.
The question is gradually changing. Instead of asking whether students have used AI, educators are becoming more interested in understanding how students have engaged with it, what decisions they have made and what evidence of learning remains visible throughout that process.
This shift was explored from multiple perspectives throughout the day, each approaching the challenge from a different disciplinary context but arriving at remarkably similar conclusions.
Robert Mitton explored the potential of Human-AI Dialogic Assessment, drawing on doctoral research into how AI can support authentic conversations between educators and learners. Rather than positioning AI as an autonomous assessor, Robert argued for a collaborative model where AI assists with transcription, evidence mapping and adaptive questioning while educators retain responsibility for interpretation, professional judgement and feedback. In doing so, he challenged delegates to think beyond AI as either a threat or a solution, instead considering how it might strengthen assessment practices by creating more opportunities for meaningful dialogue.
This emphasis on visible thinking was developed further by Jay Murphy, whose interactive workshop encouraged participants to experience AI not as a tool that provides answers, but as one that prompts deeper reflection. Through carefully designed interactions, attendees experienced first-hand how AI could be used to encourage students to explain, justify and refine their thinking rather than simply generate outputs. As Janet demonstrated, when AI refuses to provide immediate answers and instead asks better questions, the learning shifts from producing content to developing judgement.
While Robert focused on dialogue and Jay on metacognition, both challenged the prevailing assumption that assessment should primarily evaluate what students submit. Instead, they illustrated how assessment can reveal the intellectual processes that underpin learning.
Making learning visible
This theme continued throughout the afternoon.
Candace Nolan-Grant presented work from Durham University exploring process-based assessment within group projects. Rather than assessing only the final report, students were asked to demonstrate how they had collaborated, reflected, evaluated evidence and used digital tools, including AI, throughout the project. Individual screencasts provided opportunities for students to articulate their contributions and demonstrate their decision-making, helping both educators and learners make the learning journey more visible.
Interestingly, Candace reflected on an unexpected finding from the project. While some students initially questioned whether reflection was truly “academic”, markers found that these reflective narratives revealed sophisticated scientific reasoning, collaborative thinking and evaluative judgement that would otherwise have remained hidden within the final group submission. Rather than detracting from academic rigour, process-based assessment provided richer evidence of learning.
A similar philosophy underpinned Dr Catherine Hine’s work at the University of Plymouth. Through consultancy-based learning, students worked with authentic organisational challenges, using AI to navigate information and support analysis while focusing their efforts on the human capabilities that employers continue to value most. Questioning, ethical judgement, collaboration, stakeholder engagement and professional decision-making remained central to assessment, with AI positioned as a tool that supported learning rather than replaced it.
Taken together, these presentations suggested that authentic assessment is becoming less concerned with protecting traditional assessment formats and more concerned with revealing the intellectual and professional capabilities that students develop throughout the learning process.
“Perhaps the strongest message to emerge was that authentic assessment is increasingly concerned with making learning visible, not simply evaluating the product that remains at the end.”
Rethinking authenticity in an AI-enabled world
This naturally led to one of the day’s central questions: what does authentic assessment actually mean in an AI-enabled world?
In my own presentation, I reflected on the challenges encountered while developing the Authentic Assessment Framework as part of the QAA-Funded Collaborative Enhancement Project. Despite decades of literature exploring authentic assessment, the project team found that reaching a shared definition was far from straightforward. Different disciplines, institutional contexts and educational philosophies all brought different interpretations of authenticity, prompting the team to move away from searching for a single definition and instead develop a framework based around critical questions that encourage educators to interrogate their own assumptions about assessment design.
That conversation continued during the closing panel discussion, where Professor Jo Peat, Dr Paul Shore and Tina Karalis joined me to explore how authentic assessment might evolve in response to AI.
Rather than debating whether particular assessment methods were inherently authentic, the panel encouraged delegates to consider authenticity as a design philosophy. Questions about purpose, context, learner experience and graduate capability became more important than selecting any single assessment type. The discussion highlighted that authenticity is shaped not only by the task itself, but by the relationships between students, educators and the learning environment in which assessment takes place.
Across these discussions, a broader pattern began to emerge.
Whether speakers were discussing dialogic assessment, metacognition, consultancy learning, collaborative projects or authentic assessment frameworks, they were all describing different ways of making student thinking, reasoning and professional judgement more visible.
Perhaps this is where the conversation around AI is now heading.
Rather than asking how assessment can prevent students from using AI, educators are increasingly asking how assessment can better recognise the uniquely human capabilities that remain central to learning, even as AI becomes an everyday part of higher education.
We are redefining AI literacy as a graduate capability
From AI users to AI learners
While much of the public conversation around AI continues to focus on the technology itself, several presenters challenged delegates to think instead about the capabilities students need to develop alongside it.
Perhaps one of the clearest distinctions came from Dr Tim Davis of Northeastern University London, who encouraged us to consider the difference between being an AI user and becoming an AI learner. An AI user, he argued, simply uses technology to generate answers. An AI learner uses AI to deepen understanding, question assumptions and support their own learning.
This distinction feels increasingly important.
Higher education has always been about more than knowledge acquisition. It is about developing judgement, curiosity and the ability to evaluate information critically. Those capabilities become even more valuable as AI tools become increasingly capable of producing polished outputs with remarkable speed.
Tim’s practical examples demonstrated how carefully designed prompts can help students engage more critically with mathematical concepts, moving beyond simply asking AI for solutions. Although grounded in mathematics education, the principles resonated far beyond a single discipline. The challenge is not simply teaching students to use AI, but helping them understand how to think with it.
That message echoed throughout the conference.
Building capability, not dependency
A complementary perspective was offered by Dr Amani Alabed (University of Doha for Science and Technology), whose presentation explored the growing concern around AI overreliance.
Drawing on emerging research, Amani examined how frequent AI use can encourage cognitive offloading, reducing the amount of independent thinking learners undertake. Rather than presenting this as an argument against AI, however, the session focused on designing learning experiences that deliberately strengthen students’ thinking capabilities while still allowing them to benefit from AI as a learning tool.
This represented another subtle but important shift.
Rather than asking whether students should use AI, the conversation became one about ensuring AI supports intellectual development rather than replacing it. If AI removes productive struggle entirely, there is a risk that students lose opportunities to develop the critical thinking, resilience and problem-solving skills that higher education seeks to foster.
The solution, presenters suggested, is not prohibition, but thoughtful educational design.
Teaching students how to use AI responsibly
Practical approaches to AI literacy were explored further during the AI & Assessment session, led by Dr Alison Willows (University of Brighton) and Candace Nolan-Grant (Durham University).
Their work demonstrated how educators can move beyond broad institutional policies by developing assessment-specific guidance that helps students understand where AI is appropriate, where it is not, and why. Rather than relying solely on generic statements about responsible AI use, their approach encourages academics to consider each assessment individually, providing students with greater clarity while supporting transparency and assessment literacy.
This reflects a wider evolution in AI literacy.
Initially, much guidance understandably focused on institutional rules and acceptable use policies. Increasingly, however, AI literacy is becoming embedded within learning itself, helping students understand not only what AI can do, but also when its use enhances learning and when it may undermine it.
Across the conference, AI literacy was consistently presented as something far broader than technical competence. It encompasses ethical awareness, critical evaluation, digital confidence and professional judgement. In other words, AI literacy is becoming a graduate capability rather than simply another digital skill.
“The challenge is no longer teaching students how to use AI. It is helping them understand when AI enhances learning, when it limits learning and how to exercise informed professional judgement in deciding the difference.”
Looking ahead
If there was one overarching message that emerged from the conference, it was that higher education is moving beyond a reactive response to AI.
Only a short time ago, discussions centred primarily on plagiarism, detection and assessment security. Those concerns remain important, but they no longer dominate the conversation in the way they once did.
Instead, educators are increasingly asking more ambitious questions.
- How do we design assessment that values reasoning rather than reproduction?
- How do we develop graduates who can work effectively alongside AI without becoming dependent upon it?
- How do we create learning experiences that strengthen curiosity, reflection and professional judgement?
What became apparent throughout the conference was that no single institution has all the answers.
Each presenter offered a different perspective, shaped by their own disciplinary context and educational environment. Yet there was remarkable consistency in the direction of travel. Whether discussing institutional strategy, dialogic assessment, metacognition, consultancy learning, digital literacy or authentic assessment frameworks, contributors repeatedly returned to the same underlying principle. AI should not diminish learning. Instead, it should encourage us to think more carefully about what meaningful learning looks like.
Perhaps that is the most important shift of all.
Artificial intelligence is not simply prompting higher education to redesign assessment. It is encouraging us to revisit fundamental questions about why we assess, what evidence of learning we value and how we prepare graduates for increasingly complex professional environments.
Those questions are unlikely to have definitive answers.Nor should they.
The strength of the AI Literacy & Authentic Assessment in Higher Education Collaborative Enhancement Project lies precisely in its collaborative approach. By bringing together colleagues from across institutions, disciplines and professional roles, the project recognises that meaningful change will emerge through shared reflection, practical experimentation and ongoing dialogue rather than prescriptive solutions.
The conference marked an important milestone in that journey, but it is only the beginning.
Over the coming months, the project will continue with the release of two significant resources: the AI Literacy Development Tool and the Authentic Assessment Framework. Together, these resources aim to support educators as they navigate one of the most significant periods of educational change in recent decades, providing practical guidance while recognising that institutional contexts, disciplinary priorities and learner needs will continue to evolve.
If the discussions throughout the day demonstrated anything, it is that higher education is no longer asking whether AI has a place within teaching, learning and assessment.
Instead, we are beginning to ask a far more important question.
How can AI help us create richer, more authentic and more meaningful learning experiences for every student?
Useful resources
- Watch recordings from the day here.
- Connect with our speakers here.
- If you are interested in the AI Literacy Tool and Authentic Assessment Framework, register your interest and you will be notified once it is ready.
