Air Force eyeing AI-powered platform for wargaming

Anticipating the need to replenish its units during high-attrition combat with advanced adversaries like China, the U.S. Air Force is looking to industry for AI-fueled wargaming tools to play out potential scenarios.
The department issued a request for information Tuesday to apprise the deputy chief of staff for manpower, personnel and services about vendors’ capabilities.
“The Air Force plans to simulate high-intensity conflict conditions in order to stress test training and accession pipelines and explore sustainment under high attrition. An artificial-intelligence-enabled Software-as-a-Service (SaaS) wargaming platform will be used to create immersive exercises that adapt to participant decisions and generate realistic adversary actions,” officials wrote.
“AI-enabled wargaming leverages machine-learning-based red-teaming to dynamically adjust to human decisions and uses reinforcement learning and neural networks to mimic adversary decision-making; the Air Force seeks to harness these capabilities to improve readiness planning. Information obtained through this RFI will inform the Government’s acquisition strategy and future solicitation,” they added.
Gaming plays an important role in boosting U.S. military preparedness for potential scenarios, serving as “laboratories” for decision-making, analysts at the Center for Strategic and International Studies noted in commentary published last year.
However, the old-school way of conducting these types of exercises can be expensive, challenging to replicate and analytically insufficient, among other shortfalls, they noted.

AI is seen as a technology solution for those problems.
“Incorporating AI into wargames can both reduce the traditional costs associated with running games and increase opportunities for more rigorous analysis of strategy and decisionmaking,” the CSIS analysts wrote, noting that generative AI and large language models can be used to create game agents, rather than bringing in a bunch of people to sit around a table and play various roles.
“Recent studies indicate that synthetic data can effectively mirror the response patterns of a diverse array of human subpopulations, which can be useful in drawing predictive conclusions for specific sides,” they added.
The Air Force RFI noted that traditional wargames rely on static scenarios and manual adjudication.
“Recent research highlights that AI powered wargaming introduces real time adaptability, autonomous adversaries and predictive analytics, allowing scenarios to evolve in response to participant decisions. AI-driven red-teaming uses machine-learning-algorithms and neural networks to mimic adversary behavior, challenging participants to develop strategic agility. The Air Force intends to incorporate these innovations into personnel readiness exercises focused on rapid training surge and sustainment under high attrition,” officials wrote.
The department is looking for an AI platform that can dynamically inject events into exercises, adjudicate outcomes in real time, collect data on participants’ decisions, system bottlenecks and performance metrics, and synthesize insights for after-action reports.
Other desired capabilities for the platform include rapid AI-driven scenario generation; the ability to quickly adapt complex scenarios midgame based on players’ decisions; “beat system” integration to guide participants’ engagement; internal messaging capabilities that enable teams to communicate electronically and directly during gameplay; permitting teams to submit decisions, moves and actions directly within the system; options to integrate things like news reports, social media posts and videos into the game world to create a “multi-media immersive” environment; and scenario replayability, among others.
“This functionality should streamline gameplay, reduce administrative overhead, and ensure accurate data capture for adjudication and analysis,” officials wrote.
The Air Force is also interested in hosting the platform in a secure cloud environment that complies with Defense Department cybersecurity requirements, per the RFI.
The deadline for vendors to submit a capabilities statement is Aug. 26.