AI in Wargaming Simulation as a Multi-Domain Decision Support tool

Command & Control (C2) planning and decision making presents an increasingly complex challenge, due to factors such as the growing availability of relevant data and the timeframes in which this data must be processed by commanders.

By understanding likely and possible Red Force (adversary) responses to a Blue Force (friendly) commanders Course of Action (CoA), planners could be empowered to make better decisions. We propose that potential Red Force responses could be elicited through the use of Artificial Intelligence (AI) to create intelligent adversaries in wargaming platforms.

Wargaming simulation tools when used in combination with AI could rapidly predict Red Force CoA. Providing details of the current operational picture can allow AI agents to consume and process large amount of numerical and textual data such as in-game observational state. A task it is well suited to achieve.

We suggest that in future conflicts, Blue Force commanders will make effective use of decision support tools that exploit AI in wargaming simulators to find potential and novel Red Force CoAs.

This decision support tool will make use of the state-of-the-art deep learning algorithms with a multi-agent approach that allows in-game units to communicate, share experiences and collaborate towards a common operational objective. The learning process will utilise a curiosity driven reward scheme to allow the generation of novel and unexpected Red Force behaviours

The tool will allow the Blue Force commander to enter details of the current operational picture, including current force composition. A generalised intelligent adversary agent will then be allowed to be deployed to be trained on the dynamics of the current situation on massively scalable cloud infrastructure to return results at machine-speed.

The tool will then classify the generated Red Force responses into categories such as most damaging, most likely, highest risk, etc. Each of these generated CoA will have various explainability tools integrated such as trend heatmaps, virtual reality campaign playback and LLM summarisation tools to allow the Red Force behaviour to be understood at a high-level.

This will allow the Blue Force commander to make decisions faster with the anticipation of several of the possible responses Red Force could make.