When ChatGPT burst into public consciousness in late 2022, classrooms and lecture theatres were among the first places to feel its impact. Within weeks, students were experimenting with AI‑generated essays, staff were revisiting long‑standing assumptions about assessment, and universities worldwide were forced into rapid‑fire debates on plagiarism, policy and academic integrity. Amid the noise, one question loomed large: is generative AI a threat to higher education as we certainly know it, or the catalyst for a fundamental reimagining of how we teach and learn?
At UCL, that question goes beyond concern about cheating or shortcuts. Generative AI tools-capable of producing fluent text, code, images and even feedback at the click of a button-touch almost every aspect of the learning experience, from how students research and write, to how educators design curricula, assess understanding and support diverse cohorts. As global education systems grapple with these shifts, UCL is positioning itself not only to manage the immediate disruption, but to ask a deeper set of “education futures” questions: What kinds of human intelligence should universities now be cultivating? How should learning be structured in a world where data can be generated, not just retrieved? And what does “academic rigour” look like when machines can convincingly imitate it?
This article explores how UCL’s Teaching & Learning community is responding to the generative AI revolution-experimenting with new pedagogies, reshaping assessment, and engaging students as co‑designers of an AI‑rich future.Rather than taking a stance of resistance or uncritical embrace, UCL is charting a more demanding path: treating generative AI as a lens through which to rethink the purposes, practices and promises of higher education itself.
Harnessing generative AI to reimagine assessment and feedback in higher education
Rather of treating machine-generated text as a threat to traditional testing, universities are beginning to use it as a catalyst for more authentic, process-focused evaluation. Academics are designing assignments where students must critique, verify and improve AI outputs, making visible the often-hidden cognitive work of questioning, evidencing and refining ideas. This shift is also prompting greater clarity around assessment criteria: students can now compare AI-generated responses with marking rubrics,exposing what quality looks like and how judgement is applied. In seminars and labs, educators are experimenting with AI-supported simulations, scenario-based tasks and multimodal artefacts that better mirror the complexity of real-world professional practise.
Feedback is undergoing a similar transformation. Instead of receiving sparse comments weeks after submission, students can access layered responses that blend immediate AI-generated insights with targeted human guidance. For example:
- Draft analysis: instant formative feedback on structure, argument and clarity
- Evidence checks: prompts to strengthen referencing and data use
- Dialog-based reflection: AI-assisted questioning to help students articulate decisions
- Personalised action plans: tailored next steps derived from common patterns across cohorts
| Practice | Role of GenAI | Benefit for Students |
|---|---|---|
| Iterative drafts | Rapid critique of evolving work | More confident redrafting |
| Oral assessments | Live prompts and transcripts | Richer reflection and review |
| Group projects | Shared feedback workspace | Clearer roles and accountability |
Supporting academic integrity and critical AI literacy in student work
Universities are under growing pressure to distinguish between meaningful engagement with generative tools and shortcutting the learning process. At UCL, the emphasis is shifting from simply “detecting cheating” to cultivating transparent, well-documented use of AI that students can justify and critically reflect on. This means encouraging learners to keep brief usage logs, cite tools alongside traditional sources, and explain how AI outputs were evaluated, adapted or challenged. Rather than banning technology outright, educators are beginning to use its presence as a lens through which to teach methodological rigour, source criticism and the ethics of data and authorship. In many programmes, the discussion is moving from “Did you use AI?” to “How, why and with what limitations did you use it?”
Developing critical AI literacy is now seen as core to graduate attributes, not an optional add-on. Students are being asked to interrogate model bias, identify hallucinations, and understand how training data and platform business models shape what they see on screen. Staff, in turn, are revising assessment designs to reward process, originality of argument and reflective commentary. Some departments are already piloting shared frameworks like the one below to make expectations transparent and to support disciplinary nuance:
- Make AI-visible: Require brief AI usage statements or appendices.
- Value the process: Mark drafts, annotations and decision-making, not just the final product.
- Interrogate outputs: Ask students to critique AI responses against scholarly sources.
- Align with ethics: Embed discussions of privacy, consent and data governance.
| AI Use | Permitted | Conditions |
|---|---|---|
| Idea generation | Yes | Declare tool; refine ideas independently |
| Draft editing | Limited | Keep original drafts; reflect on changes |
| Full text production | No | Assessment must reflect student’s own writing |
| Fact-checking | With caution | Verify against peer‑reviewed or primary sources |
Designing inclusive, human centred learning experiences with AI tools
Rather than replacing pedagogy, generative tools can sharpen the focus on what makes learning genuinely humane: empathy, agency and belonging. When course teams intentionally foreground accessibility and equity, AI can be used to offer multiple pathways into the same intellectual space – from simplified explanations and multimodal summaries to language scaffolds for multilingual students. Educators can ask AI to generate case studies that better reflect diverse identities, disciplines and life experiences, then invite students to critique and refine them, turning bias-spotting into an explicit learning outcome. In this model,AI becomes a prompt for critical dialogue about whose knowledge is represented,whose is missing,and how technology can either amplify or disrupt those gaps.
Designing for inclusion also means giving learners meaningful control over how,when and why they use AI in their studies. Transparent signposting of AI-supported activities, clear boundaries around assessment, and shared norms on academic integrity all help students navigate these tools with confidence. Course teams can prototype activities by combining simple, structured prompts with UDL-informed design choices, such as flexible formats for engagement and expression:
- Co-created rubrics where students and staff use AI to draft, then collectively revise, success criteria.
- Accessible feedback loops that blend tutor comments with AI-generated practice questions tailored to different needs.
- Choice-based tasks allowing text,audio or visual outputs,each supported by AI exemplars for different levels of prior knowledge.
| Design focus | AI-supported practice |
|---|---|
| Accessibility | Generate alternative formats (plain language, audio scripts, captions). |
| Belonging | Localise scenarios to students’ cultures, disciplines and contexts. |
| Criticality | Ask students to audit AI outputs for bias, gaps and misrepresentation. |
Building institutional policy, staff development and governance for responsible AI adoption
Embedding generative AI into the fabric of a university demands more than isolated experiments; it requires coherent frameworks that align innovation with academic values, regulatory expectations and student rights. At UCL, this means developing clear, living documents that define acceptable use, data protection standards and expectations for academic integrity, while remaining flexible enough to evolve with a rapidly shifting technological landscape. These frameworks must be co-created with staff, students and professional services, balancing enthusiasm for new tools with critical scrutiny of bias, accessibility and environmental impact. To support implementation, governance structures need transparent lines of accountability, regular review cycles and mechanisms for rapid response when tools, laws or social expectations change.
Policy alone is insufficient without sustained investment in the people who will interpret and enact it in classrooms, labs and studios. Staff development programmes should blend critical AI literacy with practical pedagogical design, ensuring colleagues understand both how these systems work and when they should not be used. This involves layered support, from foundational briefings to advanced communities of practice, and recognition in workload models and promotion criteria. Key focus areas can include:
- Ethical and legal confidence – navigating copyright,privacy,attribution and academic misconduct.
- Pedagogical redesign – rethinking assessment, feedback and supervision in AI-rich environments.
- Technical fluency – understanding capabilities, limitations and data implications of different tools.
- Student partnership – involving learners in co-creating norms, guidance and evaluation.
| Area | Institutional Focus |
|---|---|
| Policy | Principles, risk thresholds, compliance |
| Governance | Committees, oversight, rapid review |
| Staff Development | Training, communities of practice, recognition |
| Student Voice | Consultation, feedback loops, co-design |
Wrapping Up
As generative AI continues to mature, its influence on teaching and learning will not be decided by technology alone, but by the choices educators, students and institutions make now. UCL’s work in this space points to a future in which AI is neither a shortcut nor a threat, but a catalyst for rethinking what counts as knowledge, how it is created, and who gets to participate.The next phase will demand more than new tools; it will require new literacies, new forms of collaboration and robust ethical frameworks that keep human judgement at the center. In that respect, the story of generative AI in education is still being written. The challenge for universities such as UCL is to ensure that, as these systems become embedded in everyday academic life, they serve to deepen learning, widen access and strengthen the critical capacities that higher education exists to foster.
