AI on the Battlefield: Army Plans Powerful New Decision-Making Agent for Future Wars

The South Korean Army has started the work to incorporate generative AI into its operations, with plans to build an AI system that can analyse battlefield data and make operational recommendations, officials said on 24th April. 

 

The move comes as artificial intelligence is increasingly viewed as a key asset in modern warfare, with its ability to rapidly process vast amounts of information and assist in identifying optimal strategies and tactics.

 

The Army Information and Communications School has recently announced a research tender to develop and test a prototype system for the pilot application and validation of AI agents in the defence sector, according to the military.

 

The project will centre on the development of an AI agent, a software system that will leverage generative AI (GenAI) to comprehend user goals, generate plans, and independently execute actions using external tools, while also evaluating the outcome using empirical tests.

 

The prototype will serve as a benchmark for the development of a more sophisticated system that will be able to assist commanders in analysing data, evaluating scenarios and making recommendations on various aspects of military operations, officials said.

 

From weapons systems to combat support systems to administrative tasks, “applying AI technology is required to lessen manpower, time and cost,” an Army official said.

 

To complete complex decision-making tasks, it is essential to implement AI agents that can handle unstructured data, as opposed to existing robotic process automation or simple chatbots, which can handle only structured tasks.

 

The research will also explore the use of new technologies like retrieval-augmented generation (RAG), which relies on external databases and documents to enrich the information provided by large language models, as well as the automation of repetitive tasks and the extraction and analysis of information on resource management.

 

The study focuses on the analysis of global and domestic trends of vision language models and AI agent technologies, prototype design and development of a study system and pilot applications, as well as validation tests.

 

The prototype will be used to gather tactical data and improve evaluation processes for the Army to make improvements to the system’s learning capabilities for future development.

 

The outcomes can be utilised in the development of the Korea Joint All-Domain Command and Control System (KCCS-A), the officials said, adding that it can further be used for identifying key implementation tasks and integrating intelligent AI agents into the operational process. The system can also be used as a training system for military personnel.