---

Follow the Signals: Trustworthy AI for Distributed Data and Healthy Digital Discourse

Dr. Marco Fisichella, Research Group Leader and Senior Strategy Officer, L3S Research Center of Leibniz University of Hannover, Germany

Abstract:

Trust in AI systems requires both technical reliability and societal responsibility. This talk explores how trustworthy AI can be achieved by following signals at two complementary levels. At the technical level, I present methods for learning from distributed and sensitive data through federated learning, focusing on privacy, robustness, fairness, and interpretability. At the societal level, I discuss how social signals extracted from digital discourse, such as stance, emotions, ethics, and toxicity, can be leveraged to support healthy online debates and responsible content moderation. The presentation provides a broad overview of my research agenda and results, and briefly outlines how these topics inform my teaching activities, contributing to the education of students in trustworthy and socially aware AI.