Explainable AI for Learning about Digital Disinformation (EKILED)

In the current era, characterized by digitally networked publics, disinformation poses a substantial threat to the democratic common good and to trust in the state and its democratic institutions. Recent technological advancements and the rapidly increasing availability of data in the field of digital disinformation enable Artificial Intelligence (AI) to detect such disinformation with high precision. However, due to their complexity, AI models—such as Large Language Models (LLMs) employed for disinformation detection—function as "black boxes" for users. Consequently, users are unable to comprehend or validate how and why an AI model reaches its predictions. Specifically, the underlying patterns an AI model utilizes to identify digital disinformation remain concealed.

This problem forms the foundation of the research project "Explainable AI for Learning about Digital Disinformation" (EKILED). The project aims to investigate interactive approaches to Explainable Artificial Intelligence (XAI) that disclose and process the implicit knowledge of AI models for disinformation detection in a didactically valuable manner for users. Specifically, the EKILED demonstrator is to be developed. Utilizing interactive XAI approaches, this demonstrator will serve students (the primary target group) as a reflection partner in identifying digital disinformation. The objective is to make this tool universally accessible, thereby providing educators with a low-threshold resource to teach informed and critical engagement with digital disinformation. 

The project is funded by the Federal Ministry of Research, Technology and Space (BMFTR). The objective of the BMFTR funding measure, "Trust in Democracy and the State: Detecting and Countering Digital Disinformation," is to sustainably strengthen research, development, and innovation capacities in the detection and mitigation of disinformation. It also seeks to advance effective solutions for addressing disinformation campaigns and digital manipulation. The initiative aims to expand both research expertise and societal media literacy, while actively promoting the transfer of research findings into practical applications. The EKILED project contributes to societal resilience and the strengthening of democratic structures. Furthermore, it serves as a guiding framework for Germany's technological sovereignty, particularly concerning research in the domain of trustworthy and explainable AI.

Project Details

• Collaborative Partners: University of Ulm (project coordination),  University of Bamberg, DASU – Transfer Center for Digitalization, Analytics & Data Science Ulm, Eduversum GmbH
• Project Duration: March 15, 2026 – March 2029
• Funded by the Federal Ministry of Research, Technology and Space (BMFTR)

More coming soon...