News

Prof. Dr. Ute Schmid from the University of Bamberg will give a lecture as part of the “Lecture Series on AI in Science and Society” on 9 May 2025 from 2:15–3:45 p.m. in H27 (GW II). The title of her talk is: “Methods for Trustworthy AI: Can We Transfer Lessons Learned from Classifier Learning to Generative AI?”

Abstract: With the advance of highly performant AI systems – deep learning based classifiers and transformer based generative approaches – there is hope that AI tools can support humans to master complex tasks more efficient as well as in high quality. However, especially for critical tasks such as medical diagnostics or generation of program code for scientific tasks, AI systems need to be reliable and trustworthy. In the talk I will introduce the seven trustworthiness requirements proposed by the European Union High-level Expert Group on Artificial Intelligence. I will discuss the relation between human trust in AI systems and the trustworthiness requirements. For the requirements of performance, robustness, transparency, fairness, and human agency and oversight I will present methods which have been developped in the context of classifier learning and discuss possibilities how these requirements could be addressed for generative AI tools.