User-Driven Explainable AI in Education and Its Implications for Human Agency

The rapid integration of artificial intelligence (AI) into educational contexts has intensified concerns regarding transparency, accountability, fairness, and stakeholder agency. Explainable AI (XAI) research has the potential to advance methods for interpreting complex machine learning models, revealing opportunities to utilize explainability to enable safe, sustainable, and user-driven implementation of AI-supported systems in authentic educational settings. This half-day interdisciplinary workshop addresses this potential by focusing on explainability by design as a mechanism for strengthening the agency of teachers, learners, and institutions. Bringing together researchers and practitioners from AIED, EDM, and L@S, we aim to develop a shared understanding of design principles, XAI frameworks, technical challenges, and a research agenda for embedding transparency across the entire AI life cycle, from data pipelines and modeling to classroom integration and governance. We utilize targeted discussions, engaging talks and group activities to emphasize collaborative and cross-disciplinary exchange and long-term community building. By re-framing explainability as a socio-technical design challenge rather than solely a technical feature, this workshop advances the field toward more accountable, equitable, and agency-supportive AI in education.