ECTEL 2026: Keynote Speakers

Abstract: The AI community is increasingly interested in investigating explainability to foster user acceptance and trust in AI systems. However, there is still limited understanding of the actual relationship between AI explainability, acceptance and trust, and which factors may impact this relationship. I argue that one such factor concerns individual differences among users, including long-term traits (e.g., cognitive abilities, personality, preferences) and short-term states (e.g., cognitive load, confusion, emotions). Namely, given a specific AI application, different types and forms of explanations may work best for different users and even for the same user at different times, depending to some extent on their long-term traits and short-term states. As such, our long-term goal is to develop personalized XAI tools that dynamically adapt to relevant user factors. In this talk, I will present a general methodology for this investigation and examples of how we applied it to understand the importance of personalized XAI in AI in Education

Cristina Conati is a Professor of Computer Science at the University of British Columbia, Vancouver, Canada. She received an M.Sc. in Computer Science at the University of Milan, and an M.Sc. and Ph.D. in Intelligent Systems at the University of Pittsburgh. Cristina has been researching human-centered AI and AI-driven personalization for over 25 years, with contributions in the areas of  Information Visualization, Intelligent Tutoring Systems, User Modeling, Affective Computing,and Explainable AI.

Cristina’s research has received 10 Best Paper Awards and  the Test of Time Time Award  2022 from the Educational Data Mining Society. She is a Fellow of the Association for the Advancement of AI (AAAI) and of the Association for Computing Machinery (ACM), and co-Editor in Chief of the International Journal of AI in Education. She served as President of Association for the Advancement of Affective Computing (AAAC), as well as Program or Conference Chair for several international conferences.

Abstract: Generative AI is an existential threat to the current educational system: if machines can now perform many of the tasks that schools and universities have long valued, what should education now be for? This paper argues that the answer lies not in intensifying individual performance but in redesigning education for collective intelligence. In a world marked by ecological crisis, social fragmentation, and deep technological interdependence, the central educational challenge is to cultivate our capacity to think and act well together. I see this challenge in historical perspective. Education has never been separate from technology: from the painted caves and initiation rituals of the earliest teaching, through the scribe schools of ancient Sumer, to Socrates’ anxieties about writing and Comenius’ vision of print “impressing” knowledge onto pupils’ minds, each new medium has reshaped what it means to learn. I also introduce the philosophical question ofwhether knowledge is a thing to be transferred or a relationship to be realised through dialogue. Seen this way, the retreat to “human” pedagogy against technology is a false choice, and generative AI is best understood as a new medium for thinking rather than a threat to it. Drawing on dialogic theory and on my research into teaching for collective intelligence I argue that generative AI can help, provided it is used not as a substitute for human thought but as a resource for widening inquiry, engaging difference, and supporting shared knowledge creation. I outline a design framework for AI in education for collective intelligence, distinguishing how AI can support the grouping of learners, the micro-processes of dialogue, the staging of inquiry over time, and the calibration of AI’s own agency. The aim of education becomes the design of dialogic environments in which learners, teachers, and AI systems together contribute to deeper understanding, better judgement, and more responsible collective action.

Rupert Wegerif is Professor of Education in the Faculty of Education at the University of Cambridge and the founder and academic director of the Digital Education Futures Initiative (DEFI) at Hughes Hall, Cambridge. He is the author of several influential books and articles in the area of educational theory, educational psychology and education with technology. His recent talks, articles and books offer a theoretical foundation for the design of dialogic education with AI and with technology more generally.

Abstract: As generative AI moves from experiment to classroom, our field faces a dual challenge: grounding these systems in proven learning science, and building the evidence base to know whether they improve student learning outcomes. Drawing on recent developments at Google DeepMind, this keynote addresses these challenges through a combination of intentional pedagogical design and rigorous longitudinal evaluation. We introduce pedagogical instruction following: a design approach that enables educators to shape AI tutoring behaviour according to established instructional strategies. Additionally we describe the technical and user-experience challenges of integrating this approach into Gemini via Guided Learning.

This talk presents evidence from longitudinal efficacy studies, including one conducted across UK secondary schools using LearnLM, a variant of Gemini fine-tuned specifically for pedagogy. Students in the Eedi online maths tutoring platform received either a static hint, an interaction with a human tutor, or an interaction with LearnLM supervised by human tutors. LearnLM proved to be a reliable source of pedagogical instruction, with 76.4% of suggestions accepted by human tutors without edits. Most notably, students guided by LearnLM were 5.5 percentage points more likely to independently solve novel problems than those supported by human tutors alone. These results suggest that intentional pedagogical design is a critical factor in delivering effective, individualised learning support at scale.

Katie Stasaski is a Research Scientist on the AI and Education team at Google DeepMind, where she focuses on improving and evaluating LLMs for education. Before joining Google DeepMind, she was a Senior Applied Scientist at Salesforce AI Research, building AI tools to enhance developer productivity. Katie holds a PhD from UC Berkeley, where her research centered on automatic question generation, tutoring systems, and automated feedback gemeration.