Welcome to relAI! – Konrad Zuse School of
Excellence in Reliable AI
The current technological revolution is largely driven by spectacular progress in artificial intelligence (AI). Yet, although the huge potential is widely recognized, the lack of reliability of AI technology is still considered a serious issue of concern, limiting its adoption both by industry and society at large. Indeed, aspects such as safety, security, and privacy-preservation are essential prerequisites for the use of AI in domains of public interest – e.g. ensuring that robots do not endanger life or respecting confidentiality of data.
The vision of the 'Konrad Zuse School of Excellence in Reliable AI' (relAI) is to train future generations of AI experts, who for the first time combine technical brilliance with awareness of the importance of AI’s reliability. Our novel, highly innovative AI program will educate top international candidates in the end-to-end development of reliable AI systems (including scientific knowledge, business expertise, and industrial exposure), both for industry and academia, and perform cutting-edge research to make AI ready for deployment in critical application domains.
relAI trains future generations of AI experts who combine technical brilliance with an eye on AI's implications for society.
NEWS
Can We Trust the Uncertainty of Causal Foundation Models?
Causal foundation models promise fast, flexible treatment-effect estimation from observational data, but can their uncertainty be trusted? The post of PhD Student Valentyn Melnychuk examines this question in light of the new class of foundation models for tabular and causal inference, called prior-data fitted networks (PFNs). He found that the answer is not always positive, but that … Read more
...more
relAI with SIPTA Summer School
relAI is proud to support SIPTA, the Summer School on Imprecise Probabilities 2026, which will be held in Munich from Monday, July 27, to Friday, July 31, 2026. The school is organized by the Chair of Artificial Intelligence and Machine Learning (AIML), Institute of Informatics, and AG Augustin, Department of Statistics, both at LMU Munich. The organizing … Read more
...more
Exploring XAI Methods for Interpretability of Large Language Models
Large language models (LLMs) process vast amounts of information to produce impressive outputs, such as summarizing texts, answering questions, and classifying sentiment. However, their operation often feels like a black box. 🤔 How do they make predictions? Can we identify which input information is relevant to the model’s final decision? In her blog post, relAI PhD Molly … Read more
...more