The role of the educator in a world with Generative AI: UCD/Sheffield University Practice Exchange Seminar

I was asked to speak at the UCD/Sheffield University Practice Exchange Seminar this afternoon (flyer below my remarks.) It has been a very long time since I’ve posted (primarily because I’ve been running at 200% getting up to speed in my new role and trying to keep the other plates that were in the air spinning!)

Instead of going the “slides” route, I wrote out my talk (something I haven’t done in a while) and, thought it would make sense to share my thoughts here. Not sure if the world needs more hot takes on GenAI, but, for better or for worse – here are my thoughts on the prompt that was given to me: The role of the educator in a world with Generative AI 

============

I’ve been working in education and “educational technology” since 1997. I started out as a network administrator at a secondary school and where I was also in charge of teacher professional development in the tech lab. The World Wide Web as we know it had gone public in 1993 and Google came online in 1998 – fundamentally changing the way we interact with information. Since then, I’ve lived through the promises and the panics at the intersection of technology and education: 1:1 devices, MOOCs, Second Life, Twitter/Social media, blockchain…all technologies that many claimed would “disrupt” higher education as we knew it. Each wave brought proclamations that the role of the educator was fundamentally, irreversibly changed. And here we are again in the GenAI era. 

I will admit this time does feel a bit different. It is commercially ubiquitous, intentionally subversive, and because it arrived at a moment when many of our students and faculty are exhausted, over-assessed, and asking questions about whether and how the work we do in classrooms matter.

When we talk about the title of this talk: “the role of the educator in a world with Generative AI,” I think we’re really asking a deeper question: What is education actually for?

Because if education is primarily a credentialing exercise, a process of producing rote artifacts or essays that demonstrate knowledge of content, then yes, Generative AI poses an existential problem. Anyone can now outsource that process to a tool that will produce something plausible, referenced, and stylistically competent in seconds. And no plagiarism detection tool can reliably catch it.

The HEA’s Principles for Ethical AI Adoption (2025), which I’m going to reference frequently in this talk, published just this past December, are very direct on this point: AI detection systems produce probabilistic scores, not concrete findings. They are, as the document puts it, “prone to both false positives and false negatives.” (p.14) They disproportionately flag the work of non-native English writers as AI-generated, while sophisticated AI users can often escape notice entirely. The framework goes so far as to say that detection tools are “not recommended as primary evidence in higher education.” (p.14) Full stop.

So if the essay or final artifact-as-proof-of-learning is no longer a reliable instrument, and detection is not the answer – what do we do?

Here is where I think we have to resist the pull toward quick fixes. Because the temptation is to respond to a technology problem with more technology. And technology companies and venture capitalists are more than happy to spin up or vibe code a tool to “fix” the problem.  

I believe that our fundamental role as educators has not changed. I would align my own educational practices and beliefs with philosophers like Paulo Freire and Nel Noddings. I reject a passive “banking model” of knowledge transfer. We must sit with and care for students, sometimes uncomfortably, in the space between what they know and what they don’t yet know. Teaching and learning is a combination of care, reflection and action and a practice that empowers people to deal critically with their reality and discover how to participate in the transformation of their world.

This is something the HEA guidance reinforces. It insists that decisions about AI integration “cannot be reduced to questions of efficiency or competitiveness” (p.2) they must be “framed in terms of educational purpose.” (p.2) What kind of learning experiences do we want to foster? What skills and dispositions should students develop? What forms of scholarly engagement do we want to protect?

Those are not new questions. They are the questions of critical digital pedagogy. They are, at their core, the questions of any well-designed education. The HEA guidance explicitly states that “thoughtful rejection of gen AI tools is as valid as critical integration when pedagogical judgement demands it.” (p.4) That is extremely important to surface. You are not obliged to embrace every tool that lands on your screen and you and your students can be “AI literate” without ever touching an LLM. 

Many educators are carrying the weight of the AI and AI and assessment challenge as if it were a personal, individual failing. As if the right response is to redesign every rubric and rebuild every assignment from scratch – alone. That is not sustainable. And it is not fair.

Again, the HEA guidance reinforces this: “These responsibilities cannot rest solely with individual academics; they require institutional systems that provide consistent support for assessment design, compliance, and AI literacy.” (p.8) It goes further by naming workload recognition, shared repositories of exemplars, advisory support structures, and programme-level coordination as institutional obligations, collective action – not individual acts of good faith.

Assessment design in the age of Generative AI, as the guidance puts it, is “no longer a purely pedagogical exercise, but a cross-cutting policy concern.” (p.9) It means that when you are struggling to know how to design an authentic assessment, or to handle a suspected case of AI-generated work, or to write a clear AI-use statement for your module brief, you are not failing. You are operating in a space that genuinely requires collective, institutional response. One that we must all contribute to.  

The guidance recommends approaches to assessment that make authorship and judgement visible: staged drafts, oral assessments, reflective portfolios, unique datasets, in-person work. Not because these are perfect solutions, but because they shift the focus back to what we actually care about: evidence of thinking, not evidence of output.

I am wary of the “AI is an opportunity” or “AI is inevitable” framing, as it can so easily slide into the kind of toxic positivity that washes over real and legitimate concerns. The HEA guidance frames its overriding commitment in very clear terms: “The adoption of generative AI in higher education must strengthen, not erode, the institution’s capacity to fulfil its mission as guarantor of legitimate knowledge and contributor to the public good.” (p.5) That is a values statement and it centres the mission of higher education. Not tools.  

The disruption caused by Generative AI is an invitation (albeit, maybe an unwelcome one), just as the disruptions we experienced during COVID were, not to reinvent education wholesale, but to ask with renewed urgency the questions we should have been asking all along. What are we actually assessing? Do our students understand why? Are we creating spaces where genuine learning can happen, or are we running elaborate compliance systems dressed up as pedagogy?

There is also something in the HEA guidance of what it calls “preserving foundational intellectual capabilities.” (p.40) Writing, critical analysis, reasoning, creative problem-solving (which as a side note also happen to be core competencies of Education for Sustainable Development) are not incidental to education. They are the point and there is a real risk that the convenience of AI could erode the productive struggle through which those capabilities actually develop.

Nearly thirty years into this work, I keep coming back to is this:

The cold digital zeros and ones of our technological tools do not change what teaching and learning is. Learning is human. Learning is relational. It is uncertain. It involves struggle, confusion and joy and success. No language model can replicate that, nor should we try to make one that does. 

The HEA guidance, in its most foundational principle, states that “human accountability cannot be delegated.” (p.4) The moral weight of educational decisions, decisions that shape human potential, cannot be handed off to a system that has no comprehension of the consequences.

I will end with three points that are hopefully practical as takeaways: 

  1. The “AI problem” is collective, not personal. If you are struggling with AI and assessment right now, that is not a failure of individual ingenuity, it is a sign that you are working in a space that genuinely requires collaboration, institutional infrastructure, shared design, and programme-level coordination. We all know about the silos in education, and I know it’s often easier to do our own thing, but…collaboration is key. 
  2. You are not obliged to embrace every tool. AI literacy does not require touching an LLM. Thoughtful, pedagogically-reasoned rejection is as valid as critical integration. 
  3. Protect the productive struggle. Writing, reasoning, critical analysis, these capabilities develop through difficulty, not around it. The greatest risk of AI convenience is not academic dishonesty. It is quietly (and sometimes not so quietly) eroding the very processes through which genuine learning happens.

====

UCD/SHEFFIELD UNIVERSITY PRACTICE EXCHANGE SEMINAR
The Role of the Educator:
AI can readily explain and contextualise concepts, design teaching materials, grade assignments, and offer wellbeing advice. How can educators support students to harness the benefits of such tools rather than blindly trust any output?

Following on from the successful seminar with Sheffield University in 2025, discussing the roles of increasingly sophisticated technologies in fostering creativity and once again under the umbrella of the Engineering and Architecture Teaching & Learning Blended Learning Community of Practice (CoP), we have planned another in-person Practice Exchange Event with Sheffield University, connecting remotely, on Wednesday, 11th March 2026.

UNIVERSITY COLLEGE DUBLIN | Dr. Leigh Wolf

Title: The Role of the Educator in a World with Generative AI

As Generative AI tools capable of explaining concepts, generating materials, and even providing feedback flood the market, the role of the educator is shifting in ways we are only beginning to understand. This presentation will explore how educators can engage critically in a world where commercially available Generative AI exists and will persist. It will also address the very real challenges AI poses for assessment and authentication of student work – challenges that no individual educator can solve alone. The role of the educator, therefore, extends to continuing to work collaboratively with colleagues, programme and module coordinators to develop institutional approaches that are both practical and pedagogically sound.

SHEFFIELD UNIVERSITY | Dr. Eleni Routoula and Prof. Mo Zandi

Title: The Role of the Educator in a World with Generative AI (Sheffield perspective)

As Generative AI becomes embedded in students’ learning practices, educators are increasingly required to move from control towards critical engagement and meaningful integration. This presentation will share emerging practice from the University of Sheffield, including the use of AI guidelines and declarations, and curriculum activities designed to support critical engagement with AI outputs. It will reflect on implications for teaching, assessment, and academic integrity, and consider how educators can support ethical and effective AI use in line with evolving expectations from industry.

Join us for an engaging discussion and Q&A!