The goal of the ELLIS Unit Tübingen is to build a novel public research institution to attract the best scientists to advance AI, train top international students, and generate positive impact in science and society. The research agenda of the unit aims at building learning systems that approach the versatility and robustness exhibited by natural intelligent systems. Machine learning (ML) is at the heart of a technological and societal revolution, yet today’s learning systems do not generalize well to new situations, cannot learn from few examples, and do not infer causal relationships. Addressing these deficits and developing robust AI systems will be key to efficient robot teaching and explainable AI and thus help ensure technological leadership and deploying AI systems responsibly and to the benefits of society. Furthermore, the ELLIS Unit actively supports its members to tackle highly visible research challenges and to engage in start-ups, part-time employment, and other activities that help society benefit from AI research.