The US Army is looking for help from AI to predict what its enemies will do minutes or even hours before they do it

• The US Army wants help with "continuous, real-time predictive visualization" of enemy actions.
• The project is spurred by fears that human analysts won't be able to keep up with complex warfare.

The US Army wants AI that can predict what the enemy will do just minutes before the enemy actually does it.

The goal is a system that can rapidly analyze massive amounts of data — far more than humans can absorb — to forecast enemy actions and continuously update that forecast as adversaries change their tactics.

The project, titled "Real-Time Threat Forecasting," calls for "continuous, real-time predictive visualization of how the threat situation could evolve over the next few minutes to hours," according to a recent Request for Information seeking industry input.

The system will almost certainly be AI-based. Among the requirements is the "ability to run feasible software on a standard laptop and take advantage of emerging artificial intelligence and/or machine learning technologies."

The project is spurred by fears that human intelligence analysts, who are expected to assess an enemy's activity and forecast their courses of action, will not be able to keep up with the growing complexity of warfare.

The Army foresees a "hyperactive" battlefield buzzing with "robotics and autonomous systems, loitering intelligent munitions, thousands of semi-autonomous entities, short-range point defenses, AI/ML-enabled multi-INT deceptions, small, distributed Soldier-Machine Teams, and self-organizing intelligence networks," according to the RFI.

To survive, armies will have to constantly adapt. The RFI points to the 2020 Nagorno-Karabakh War, in which Armenia's Soviet-era tanks and artillery were devastated by Azerbaijan's Israeli- and Turkish-made drones, and to the war in Ukraine, which has seen a continuous evolution in drone technology and tactics.

For example, the success of Ukrainian counter-drone systems prompted Russia to avoid these defenses by sending large concentrations of "kamikaze" drones at targets in Ukraine's lightly defended rear areas. Ukraine responded with mobile anti-drone teams to track and intercept the drones.

This means that the US can no longer count on the enemy sticking to the same playbook. Military intelligence can't "assume an enemy whose behavior can be modeled via a doctrinal template," the Army RFI says.

At the same time, intelligence officers are being flooded with intelligence from around the globe and gathered from across all domains using air, land, sea, and space sensors as well as human sources. There is already far more drone imagery being collected than analysts have time and stamina to sift through and analyze without the aid of AI.

"There is simply not enough capacity to perform a wide range of tasks that have become critical to how militaries currently operate," warned a 2021 report on the future of military intelligence by Britain's Royal United Services Institute think tank.

Interestingly, the US Army sees video-game technologies as key to this predictive system. Results would be displayed in a "continuously updated and projected threat picture preferably using 3D visualization technologies and techniques." Analysts could also "run simulations of how the enemy situation might progress in the next few tens of minutes to few hours."

Commercially produced video games would also be easier to use. US Defense Department-developed "interfaces often took months to learn whereas the gaming community can typically assimilate a new game within a matter of hours or days," the RFI noted.

The idea of automating human staff officers has been around for years. Some routine work could be done by machines, yet it may be easier to imagine than to implement, as some embarrassing fumbles by ChatGPT have shown. For example, low-level military intelligence staffs may not have access to all the data that AI would need, according to the RUSI report.

The bigger question is whether warfare is simply becoming too much for the human mind to cope with. Perhaps AI can predict future enemy tactics, but it may end up devising friendly ones.

Michael Peck is a defense writer whose work has appeared in Forbes, Defense News, Foreign Policy magazine, and other publications. He holds a master's in political science. Follow him on Twitter and LinkedIn.