DARPA MAGICS seeks paradigm shift in ‘predicting collective human behavior’

How many applications could be derived from DARPA’s precrime technology that seeks to understand & predict human behavior on a collective scale? perspective

DARPA is putting together a research opportunity called MAGICS that seeks new paradigms for modeling and predicting collective human behavior.

Today, the US Defense Advanced Research Projects Agency (DARPA) issued an Advanced Research Concepts (ARC) opportunity for a new program called “Methodological Advancements for Generalizable Insights into Complex Systems (MAGICS)” that seeks “new methods and paradigms for modeling collective human behavior.”

Nowhere in the MAGICS description does it mention modeling or predicting the behavior of “adversaries” as is DARPA’s custom.

Instead, it talks at length about “modeling human systems,” along with anticipating, predicting, understanding, and forecasting “collective human behavior” and “complex social phenomena” derived from “sociotechnical data sets.”

“The MAGICS ARC calls for paradigm-shifting approaches for modeling complex, dynamic systems for predicting collective human behavior”

DARPA, MAGICS ARC, April 2025

Where machine learning has failed in accurately predicting collective human behavior by scouring social media posts, purchasing patterns, and traffic dynamics, the Pentagon’s research and development funding arm is looking for totally new ways to model and predict complex social phenomena.

According to the special notice, “The limitations of current methods for modeling human systems have revealed fundamental constraints on the ability to model and forecast human behavior in complex systems, and addressing these challenges requires overcoming several significant challenges that large data sets and ML [machine learning] do not address.”

Therefore, MAGICS is looking for “entirely new methodologies, metrics, tools and theoretical frameworks that can more accurately and reliably forecast human behavior by addressing the limitations and complexities that current methods struggle to capture.”

“This ARC opportunity is soliciting ideas to explore the question: Are there new methods and paradigms for modeling collective human behavior capable of overcoming limits of statistical approaches to accurately predict complex social phenomena and capture the dynamics of evolving, open, time varying, recursive, reactive, non-ergodic systems?

DARPA, MAGICS ARC, April 2025

To get an idea of what DARPA is looking for specifically, there are five main areas of interest:

1. Data inference boundaries and limitations:
   • There is a lack of systematic methods to determine the limits of inference for a given source of data, making it unclear how the granularity and sparsity of data relates to the predictive precision it can support, and hindering the ability to establish boundaries or what can be reliably inferred from the data.
2. Alignment validation limitations:
   • There is a lack of effective methods to quantify the alignment between observable indicators and latent constructs, hindering model validity across different settings, populations, and time spans.
3. Adaptation limitations and model obsolescence:
   • Models fail to generalize across contexts due to unclear patterns of evolution, degradation, or strengthening of underlying relationships; this failure limits the ability to assess the extent to which generalization is possible. Additionally, models fail to account for changing conditions or new information, further resulting in decreased predictive performance.
4. Psychosocial domain limitations:
   • Current methods lack integration across multiple psychological theories, limiting their applicability to particular constructs, populations and contexts. These siloed approaches hinder the understanding of complex interrelationships among latent variables, from attitude formation, to identity development.
5. Complex phenomena:
   • Current methods struggle to capture the complexity and reflexivity of social phenomena (i.e., systems often change as a result of being studied), highlighting the need for new approaches to better understand and model highly complex systems.

Proposals within the scope that DARPA is looking for include those that “incorporate theoretical, computational, and empirical efforts by integrating disciplines such as psychometrics, behavioral science, data science, and machine learning.”

“DARPA is interested in developing new capabilities to enable national security decisionmakers to optimize strategies for deterring or incentivizing actions by adversaries”

DARPA, “Theory of Mind” Special Notice, December 2024

DARPA has long been in the business of predicting, modeling, and manipulating human behavior.

In December, DARPA announced its algorithmic “Theory of Mind” program, which was later named “Kallisti” with the goal of developing “new capabilities to enable national security decisionmakers to optimize strategies for deterring or incentivizing actions by adversaries.”

According to the Theory of Mind special notice, “The goal of an upcoming program will be to develop an algorithmic theory of mind to model adversaries’ situational awareness and predict future behavior.”

“The program will seek not only to understand an actor’s current strategy but also to find a decomposed version of the strategy into relevant basis vectors to track strategy changes under non-stationary assumptions.”

The author of the Theory of Mind, (now Kallisti) special notice was Eric Davis, who joined DARPA in February, 2024.

Previously, Davis was the principal scientist of artificial intelligence, machine learning, and human-machine teaming at Galois, a tech R&D company whose clients include DARPA, the US Intelligence Community, and NASA, and whose partners include the Bill and Melinda Gates Foundation.

While a program manager has not yet been name for the MAGICS ARC opportunity, Dr. Davis would seem a likely candidate given the similarities between Kallisti and MAGICS in seeking to understand and predict future human behavior, but that remains to be seen.

MAGICS and Kallisti follow a long line of DARPA research programs to monitor, predict, and modify human behavior going back decades.

“The goal of the Total Information Awareness (TIA) program is to revolutionize the ability of the United States to detect, classify and identify foreign terrorists – and decipher their plans – and thereby enable the US to take timely action to successfully preempt and defeat terrorist acts”

DARPA, Total Information Awareness (TIA), July 2002

Following the attacks on September 11, 2001 DARPA announced its now-defunct (or potentially splintered) “Total Information Awareness (TIA) program in July, 2002 “to revolutionize the ability of the United States to detect, classify and identify foreign terrorists – and decipher their plans – and thereby enable the US to take timely action to successfully preempt and defeat terrorist acts.”

Investigative journalist, author, and contributing editor at Unlimited Hangout Whitney Webb described TIA as a “precrime approach to combatting terrorism” with the purpose of developing “an ‘all-seeing’ military-surveillance apparatus.”

In 2003, the American Civil Liberties Union (ACLU) described TIA as possibly being “the closest thing to a true ‘Big Brother’ program that has ever been seriously contemplated in the United States.”

According to the Electronic Privacy Information Center (EPIC), “A key component of the TIA project was to develop data-mining or knowledge discovery tools that would sort through the massive amounts of information to find patterns and associations.”

“Ultimately, the goal is to spur the development of novel, rigorous methods and frameworks that advance our collective capacity to understand and predict human behavior with greater accuracy and nuance”

DARPA, MAGICS ARC, April 2025

DARPA research programs are designed exclusively for national defense purposes.

However, the tech, tools, and techniques developed at DARPA have a way of becoming commercial after having gone through military use.

For example, the internet, GPS, and voice assistants like Alexa and Siri all came from DARPA research programs.

If MAGICS were to become a fully-funded and successful research program, how many different applications could be derived from this type of precrime technology that seeks to understand and predict human behavior on a collective scale?

In the hands of law enforcement, how easy would it be to profile an entire population and anticipate its every move?

How would insurance companies leverage these newfound insights?

What would big tech companies do with this power? Create better ads? Nudge you in certain directions? Influence the outcomes of elections?

Why is DARPA focusing broadly on targeting “collective human behavior” and not specifically on “adversarial behavior” in its current documentation?

Are there any guarantees that the technologies and methodologies coming out of MAGICS won’t ever be unleashed on private citizens or groups in a domestic setting?

---

Image Source: AI-generated by GROK