Symposia

Behind the scenes: How students use AI (and what we can learn from it)

Sophia Mavridi: De Montford University

Despite the growing interest and debate around AI in education, discussions still tend to centre on the perspectives of teachers and decision makers, leaving student perspectives largely underexplored. This keynote draws on a phenomenographic study with L2 students across UK higher education to examine how they make sense of AI-assisted writing on their degrees. The first part will briefly outline the key findings, focusing on what students attend to when using AI and why. The second part will focus on what we can learn from these experiences as educators. I will share four research-informed takeaways that challenge some dominant assumptions about AI in academic writing, particularly around language, efficiency, agency, and academic integrity. The session will invite language educators to reflect on how these insights might inform writing pedagogy, feedback, and assessment.

Sophia Mavridi is a Senior Lecturer in Educational Technology and TESOL at De Montfort University (UK). She specialises in digital pedagogy, online learning, and AI in education. She has extensive experience in language teacher education, working with organisations such as the British Council, NILE, and Bell Foundation, and is committed to bridging research and practice.

Linkedin https://www.linkedin.com/in/sophiamavridi/ 

When Students Don’t Know What to Ask AI: Teaching Collaborative Interaction with AI Agents

Jinyang Song: Xi’an Jiaotong-Liverpool University

As AI agents become embedded in language learning, many students struggle not with access to technology, but with knowing how to meaningfully interact with it. Classroom observations from an English for Academic Purposes (EAP) course at a Sino-foreign university revealed that students often treated AI as a task-completion tool, lacking the skills to ask process-oriented, reflective, or critical questions.

To address this, an AI agent was designed specifically to teach students how to work collaboratively with other AI agents. Rather than providing answers, the agent modelled questioning strategies, prompted reflection, and scaffolded interaction, positioning AI as a learning partner within a collaborative learning framework. The agent was iteratively tested in classroom activities focusing on academic reading and writing.

Observations suggest increased student engagement, more purposeful AI interaction, and greater willingness to critically evaluate AI-generated responses. For BALEAP practitioners, this approach highlights how collaborative learning can be extended to include AI agents through explicit pedagogy, helping students develop shared norms for responsible, reflective, and effective AI use in EAP classrooms.

Jinyang (Sam) Song is a Fellow of the Higher Education Academy (UK) and a BALEAP Associate Fellow. He teaches English for Academic Purposes at Xi’an Jiaotong-Liverpool University, where he also serves as Deputy Module Leader at the Global Cultures and Languages Hub. His work focuses on the pedagogical integration of artificial intelligence in language teaching, particularly the use of AI agents to support collaborative learning, learner autonomy, and critical engagement with technology in English-medium higher education contexts.

Designing AI-Supported Tasks That Build Confidence and Interaction.

Shuangxin Zhang and Jiashi Wang: Xi’an Jiaotong-Liverpool University

The rapid mainstreaming of generative AI is reshaping digital literacy in EAP, raising a practical question: how can GenAI be used to increase interaction and strengthen learners’ critical evaluation, rather than replace thinking and communication? This 10-minute talk shares three classroom-ready designs from a Year 1 EMI EAP module (CEFR B2+) at a Sino-British university. First, teacher-built HTML materials support differentiated practice and structured peer discussion, enabling students to make informed choices about challenge and scaffolding. Second, a teacher-trained Doubao assistant delivers micro-tasks based on teacher-written prompts to scaffold student interaction in speaking practice and team planning, strengthening negotiation language and shared responsibility. Finally, an institutional XIPU AI chatbot supports writing exam preparation by scoring drafts against module descriptors, generating targeted revision advice, and recording engagement via university accounts. Taken together, these designs operationalise AI literacy as teachable classroom practices, including evaluating AI output, making informed choices, and revising in response to feedback. The talk concludes by summarising practical principles for developing AI literacy through interactive, classroom-ready EAP tasks.

Shuangxin Zhang is a Language Lecturer at Xi’an Jiaotong-Liverpool University, teaching English for over 10 years. She holds a MA in TESOL, CELTA, FHEA and the BALEAP TEAP Fellowship. Her research interest is students’ acquisition and development of the second/foreign language in EAP class and the application of AI in EAP teaching. 

Jiashi Wang is a Language Lecturer at Xi’an Jiaotong-Liverpool University. She has over 10 years of teaching experience, especially in English academic writing. She has been passionate about English teaching and intellectual enlightenment, taking an active part in researching/experimenting on reading-guided writing, peer evaluation and continuous assessment.

Making AI Evaluation Social: Group Discussion as Critical Digital Literacy Practice in EAP.

Jolanta Hudson University of Glasgow

Generative AI is now part of many students’ academic writing workflows, yet EAP responses often default to policy warnings or individual “prompt skills”. This short talk argues for a different starting point: making AI evaluation social. Drawing on focus group data from an EAP writing context (Groups 1–4), I show how students’ Critical Digital Literacy is enacted through collective sense-making as much as through individual verification.

Across groups, students described GenAI as helpful but limited: useful for writing initiation, outlining, summarising and language refinement, yet criticised for generic reasoning, weak disciplinary fit and unreliable citations. Crucially, the group discussion itself became a learning mechanism. As students compared outputs and strategies, they developed shared language for judging credibility, surfaced tacit norms about what counts as legitimate academic work, and negotiated boundaries between support and substitution. Talk turned general caution into actionable practices: constraining inputs, separating stylistic editing from claim checking, and treating AI text as provisional until independently verified.

I conclude with practical, discussion-based classroom approaches that create space for open and honest dialogue, and that help students practise Critical Digital Literacy in ways that transfer to independent EAP writing.

Jolanta Hudson is a lecturer at the University of Glasgow, where she teaches the English for Academic Purposes in-sessional and pre-sessional courses, and the Masters TESOL teacher training modules. She also supervises TESOL dissertations. In addition, she provides support and staff training on the use of educational technologies.

Her research interests focus on technology in EAP, Digital Literacies, and Intercultural Communication in academia. She is a doctoral researcher currently undertaking her PhD in E-Research and Technology-Enhanced Learning at Lancaster University.

Validity, Accessibility, and Authenticity: What EAP Can Teach the University About Assessment

Kelly Webb-Davies: University of Oxford

This talk examines how an assessment redesign in English for Academic Purposes (EAP) became the foundation for a broader framework for written assessment in higher education.

The increasing use of generative AI by students has created challenges in EAP assessment: polished written submissions are difficult to interpret as evidence of students’ academic capabilities, while attempts to restrict or detect AI undermine both validity and equity. In response, I redesigned an assessment to improve validity by capturing students’ ideas and reasoning under secure conditions, followed by allowing refinement of those ideas with no restrictions on AI use.

That EAP-specific redesign is now being developed into a more generalisable written assessment model. The emerging model addresses three core challenges: ensuring validity by securing evidence of the intended construct, embedding linguistic and neurological accessibility into assessment design, and recognising the authenticity of contemporary, technology-mediated writing processes.

The talk concludes by arguing that EAP’s disciplinary expertise in navigating academic language, academic skills, and technology-mediated writing in high-stakes contexts offers valuable insights for assessment across disciplines, especially as technology-mediated writing is no longer a niche concern, but a shared condition of learning across the university.

Kelly is the Lead Education AI Consultant at the University of Oxford’s AI Competency Centre, with a background in linguistics and EAP. She holds degrees from the University of Western Australia and the University of Melbourne, and previously taught Academic English at Bangor University. Her work focuses on the thoughtful and effective integration of AI into higher education to enhance communication, reduce bias, and expand access to knowledge. She is particularly interested in how AI can support students facing linguistic barriers and neurodivergence, and in designing pedagogy and assessment that foreground validity, inclusion, and human judgement alongside emerging technologies.

Teaching Together in Fragmented Digital Spaces: What Student Use of Technology Tells Us About Engagement and Belonging

Garth Elzerman and Ediyanto Liu: Xi’an Jiaotong-Liverpool University

Technology is now a routine part of EAP teaching, but it is not always clear what student use of digital tools tells us about learning. This short presentation examines student engagement with a range of technologies across one semester of an EAP course and asks a simple question: what happens to learning when it becomes increasingly individual and optional?

Using data from diagnostic tests, LMS activity logs, self-study tools, review quizzes, and AI-supported speaking practice, the presentation identifies a clear pattern. Early in the semester, most students engage with shared digital activities. As the course progresses, participation narrows, with smaller groups of students using optional tools, often close to assessment points.

Rather than interpreting this solely as a motivation issue, the presentation offers an alternative perspective. As learning becomes distributed across multiple platforms, students may lose a sense of learning together. Digital tools can support practice and access, but they do not automatically create shared learning experiences.

The presentation argues that teachers play a central role in restoring togetherness through deliberate choices about when to use technology and when to step away from it, supporting learning as a collective, not purely individual, process.

Garth Elzerman is a university lecturer in English for Academic Purposes with research interests at the intersection of philosophy, pedagogy, and educational technology. He is currently pursuing a PhD in Philosophy, exploring questions related to agency, responsibility, and moral psychology. His pedagogic research focuses on academic literacies, assessment design, and the critical integration of digital and AI-supported tools in language teaching. He is particularly interested in how data, analytics, and technology shape teaching practice, student engagement, and professional judgement in higher education contexts.

Togetherness Under Pressure: Digital Saturation, Attention, and Human Connection in Academic English Teaching

Elhadj Moussa BenMoussa

Technology is now deeply embedded in Academic English and EAP teaching across UK higher education, shaping how students read, write, collaborate, and are assessed. While digital tools have enabled flexibility and access, recent research points to rising levels of digital fatigue, fragmented attention, and reduced relational engagement among students and teachers (OECD, 2023; Wiederhold, 2024). This short talk explores a growing tension in EAP classrooms: how technology intended to enhance learning can, at times, undermine togetherness.

Drawing on practitioner evidence from postgraduate and Academic English teaching, the presentation examines how students navigate multiple platforms, constant notifications, and assessment technologies alongside the cognitive demands of academic literacy. It brings this practice-based perspective into dialogue with emerging scholarship on attention, digital overload, and pedagogies of slowness in higher education (Gourlay, 2021; Biesta, 2022).

Rather than advocating for either increased technology use or digital withdrawal, the talk proposes a relational threshold approach: identifying moments where technology meaningfully supports learning and moments where intentionally reducing or pausing digital mediation strengthens participation, language development, and peer connection. Practical examples include device-free collaborative tasks, low-tech drafting stages, and assessment practices that prioritise dialogue over dashboards.

The session argues that digital literacy in EAP should include not only competence with tools, but also the capacity to disconnect deliberately in order to sustain attention, inclusion, and academic togetherness.

Dr ElHadj Moussa BenMoussa is a Senior Postgraduate Research Adviser and Doctoral Development Manager at the University of East London. He works closely with postgraduate and EAP students on academic writing, research ethics, and inclusive pedagogies in digitally mediated learning environments. His interests include technology-enhanced learning, researcher development, and the impact of digital saturation on attention, wellbeing, and academic community in higher education.