Bethgelab

AI Research Group at the University of Tübingen. Mission: Human learning is rich in feats we have yet to fully mimic in machine learning—such as open-ended knowledge acquisition or cognitive mapping of the environment and actions for self-logging, navigation, reflection and planning. Our mission is to develop agentic systems that can learn, adapt, and generalize over time, mirroring the open-ended nature of learning both by individual humans and collectively in science. Our approach is rooted in data-centric machine learning, focusing on open-ended evaluation and scalable compositional learning. To this end, we explore multi-modal foundation models that support rapid retrieval, reuse, and compositional integration of selected knowledge, enabling scalable and flexible learning