Keynote speakers

From Model Behaviour to Human Understanding: The Case for Actionable Explainable AI 

As deep learning systems grow more powerful yet more opaque, explainability becomes essential not just for trust, but for accountability, scientific discovery, and regulatory compliance. This keynote traces the evolution of XAI across levels of abstraction, from low-level attributions to human-readable rationales and counterfactuals, and confronts hard truths about evaluation, bias, and the limits of current methods. The talk closes with a call to make interpretability genuinely actionable, guiding model steering and supporting trustworthy AI deployment in high-stakes domains. 

Prof. Dr. Vera Schmitt is a professor at Johannes Gutenberg University Mainz and TU Berlin and founding head of the XplaiNLP research group, working at the intersection of NLP, Explainable AI, and Human-AI Interaction. Her research focuses on trustworthy, actionable AI for high-stakes domains such as fact-checking and clinical decision-making.