Military Researchers asking industry to capitalize on machine learning for modeling complex human behavior
ARLINGTON, Va. – U.S. military researchers are approaching industry for new ways of modeling complex, dynamic systems for predicting collective human behavior that overcome challenges that so far have frustrated attempts at emulating evolving time-varying systems.
Officials of the U.S. Defense Advanced Research Projects Agency (DARPA) in Arlington, Va., issued an advanced research concepts opportunity (DARPA-EA-25-02-05) on Tuesday for the Methodological Advancements for Generalizable Insights into Complex Systems (MAGICS) project.
MAGICS seeks to use machine learning to model complex, dynamic systems for predicting collective human behavior, with a goal of spurring development of rigorous methods to understand and predict human behavior with accuracy and nuance.
For the past decade or more, there has been an assumption and hope that the explosion of digital data streams and powerful machine-learning tools would enable better understanding of how large-scale systems respond to changes, DARPA researchers explain.
Stalled progress
Progress has stalled, however, because current statistical methods are unable to create models that remain valid when applied to evolving time-varying systems. A solution requires overcoming challenges that large data sets and machine learning do not address.
The core question of the MAGICS project is this: are there new methods and paradigms for modeling collective human behavior?
Areas of interest include data inference boundaries and limitations; alignment validation limitations; adaptation limitations and model obsolescence; psychosocial domain limitations; and complex phenomena.
Data inference boundaries and limitations seeks ways to determine the limits of a given source of data, and develop the ability to establish boundaries or what reliably can be inferred from the data.
Observable indicators
Alignment validation limitations seeks to quantify the alignment between observable indicators and latent constructs, which until now have hindered model validity across different settings, populations, and time spans.
Adaptation limitations and model obsolescence seeks the ability to assess the extent to which generalization is possible, and account for changing conditions or new information.
Psychosocial domain limitations seeks to overcome siloed approaches that have hindered the understanding of complex interrelationships among latent variables, from attitude formation to identity development.
Complex phenomena, finally, seeks to find new approaches to understand and model complex systems, and work with open-world systems and data sets.
Submitting abstracts
Companies interested should submit abstracts no later than 10 July 2025 online to https://baa.darpa.mil. Companies submitting promising abstracts may be invited to give oral presentations and submit formal proposals.
Email questions or concerns to [email protected]. More information is online at https://sam.gov/opp/093a0c0386e6416ea3f48a7aea5b0047/view.
John Keller | Editor-in-Chief
John Keller is the Editor-in-Chief, Military & Aerospace Electronics Magazine--provides extensive coverage and analysis of enabling electronics and optoelectronic technologies in military, space and commercial aviation applications. John has been a member of the Military & Aerospace Electronics staff since 1989 and chief editor since 1995.
