On Saturday, November 22, TRR 318 presented a diverse program at the Wissenswerkstadt Bielefeld during the GENIALE 2025 science festival. Researchers from the fields of psychology, linguistics, computer science, and sociology of technology provided insights into their work on the explainability of artificial intelligence at several stations. Visitors had the opportunity to learn about key issues addressed by the research network and try out various AI applications for themselves.
One component of the program was the “Wissenshappen”(“Knowledge bites”): Dr. Angela Grimminger, Prof. Dr. Heike Buhl, and Prof. Dr.-Ing. Britta Wrede explained fundamental aspects of explanation in short sessions—both in interpersonal communication and in the context of AI systems. The format offered a compact overview of current research perspectives in the TRR and was used by participants for in-depth discussions.
At the same time, workshop participants were able to learn how image-processing AI systems make decisions and why they sometimes deliver unexpected or incorrect results. Typical challenges emerged, such as the sensitivity of the models to irrelevant image properties. This illustrated how central traceable decision-making processes are to the development of trustworthy AI.
Helen Beierling from the B05 project also presented a robotic arm that is used in her studies to teach test subjects how to play mini golf. Augmented reality glasses are used to visualize planned movements, thereby contributing to a better understanding of AI processes.
At the C06 project exhibit, Jana Wiechmann and Frederik Rautenberg gave visitors an insight into the production of voice characteristics using AI. The demonstration showed how differently voices sound when individual acoustic characteristics, such as roughness, are changed.
Overall, the afternoon offered a wide range of approaches to AI research in TRR 318 and gave visitors of all ages a practical insight into the topic of explainability.
