Bridging Prediction and Intervention Problems in Social Systems

Machine learning models are often assessed by the quality of their predictions, yet their real-world impact extends far beyond these metrics. Models function as interventions within complex social systems, influencing stakeholders, infrastructure, and decision-making processes. This workshop challenges the prediction-centric paradigm, bringing together an interdisciplinary community to explore alternative frameworks that integrate statistical modeling and tools, policy intervention design, and evaluation methodologies. By shifting toward an intervention-based perspective, we aim to foster new methods and theories that bridge machine learning and real-world deployment in social systems.