Generative Intelligence Lab

The main limitation of current artificial intelligence (AI) systems is that they do not have a causal generative understanding of our world. We believe that this deficiency is the root cause of foundational open problems in AI, such as the need for large-scale human supervision, the lack of interpretability and the lack of safety and security. Therefore, we integrate aspects from Machine Learning, Visual Computing, and Computer Security to develop generative AI systems that learn self-supervised while being a safe, secure, and human understandable component of our everyday lives.