FDW Causal Prediction of Human System Dynamics
VT-ARC collaborated with OUSD(R&E)/BRO to organize the Future Directions Workshop on Causal Prediction of Human System Dynamics. VT-ARC identified leading academic researchers in modeling, data science, and human behavior to co-chair the event. Using open-source data analytics, VT-ARC and the co-chairs curated a diverse, interdisciplinary participant list spanning the social sciences, engineering, and computational modeling. The workshop was held in Stanford, California, and the final group included researchers and government observers with expertise in complex systems, machine learning, causal inference, and behavioral prediction.
The workshop explored how emerging advances in causal modeling, dynamic systems, and machine learning can be harnessed to predict and understand human behavior at scale, especially in complex, high-stakes environments relevant to national security. Participants emphasized moving beyond correlation-based models toward frameworks that support intervention-aware, generalizable predictions of human-system dynamics. Key research areas included development of formal causal frameworks for multi-agent systems, integration of computational social science with statistical learning, and construction of hybrid models that blend data-driven and theory-informed approaches. The report highlights the need for new infrastructure, including testbeds and standardized datasets, to support empirical validation of causal models and cross-disciplinary collaboration. This work lays the foundation for predictive tools that are both scientifically rigorous and operationally actionable in contested and rapidly evolving human environments.
