Causal machine learning is rapidly emerging as a cornerstone for the next generation of scientific discovery. While traditional machine learning excels at pattern recognition, it often struggles to address questions of *why* phenomena occur and *how* interventions will change outcomes. By integrating causal reasoning with advances in representation learning and foundation models, we can move beyond correlation to build systems that support robust scientific understanding, reliable decision-making, and actionable knowledge (Feuerriegel et al, 2024, Nature Medicine).

Objectives

By combining the different strengths from both groups, we can:
– Identify synergies across projects and disciplines, enabling richer collaborations.
– Share best practices for designing, training, and evaluating causal models in scientific contexts.
– Map out strategic research directions where foundation models can be extended with causal principles to ensure interpretability, generalization, and fairness.

For us, this focus is particularly important: our work depends on models that do not only predict but also explain and guide interventions. Developing a shared vision around causal ML ensures that our teams remain at the forefront of trustworthy AI in science, while fostering a culture of cooperation and knowledge exchange.

Why now? 

Importantly, our groups approach causal ML from different but complementary angles. Stefan Bauer’s research emphasizes causal discovery and representation learning in biology, aiming to uncover underlying mechanisms and abstractions. Stefan Feuerriegel’s group focuses on treatment effect estimation and decision support in medicine, targeting questions of interventions, outcomes, and clinical practice. These perspectives involve distinct methodological traditions—causal discovery vs. causal inference—that now need to be reconciled. With the advent of foundation models, we have a unique opportunity to align these approaches, develop shared methodologies, and create systems that are at once explanatory, predictive, and prescriptive.

Please note: This is just an information regarding events taking place at SSC; public attendance is therefore not possible due to the character of the retreat.