SODA @ ICML 2025: Papers and Awards

SODA Lab at the 42nd International Conference on Machine Learning (ICML)

The SODA Lab was prominently represented at the 42nd ICML, one of the largest and most prestigious conferences in machine learning, attended by over 9,000 researchers from around the world. This year’s contributions focused on bridging technical machine learning research with institutional and societal challenges, especially in the public sector.

Designing AI Systems that Adapt to Societal Change
Prof. Dr. Frauke Kreuter delivered a keynote (https://icml.cc/virtual/2025/invited-talk/39871) on adaptive alignment, proposing strategies for aligning AI systems with evolving societal norms. Drawing on tools from survey research and value studies, the talk called for closer collaboration between machine learning, statistics, and the social sciences.

When Is Better Prediction Worth It?
Unai Fischer Abaigar, Prof. Dr. Christoph Kern, and Juan Carlos Perdomo (Harvard, NYU) received one of six Outstanding Paper Awards for their work “The Value of Prediction in Identifying the Worst-Off” (https://arxiv.org/abs/2501.19334).