Analysis on the Explainable of Generative Artificial Intelligence in Military Contexts

The “black box” characteristics of generative artificial intelligence (generative AI), which is based on large language models(LLMs), contradict the need for decision certainty in military contexts. This paper explores the technical application and potential development of explainable generative AI in military contexts. It starts by introducing the technology development and military application field of generative artificial intelligence, followed by presenting a framework with three main components—input, interpretation, and feedback—to achieve explainability in generative AI. Finally, the paper demonstrates the application of explainable generative AI in the military field using two examples: target strike selection and a visual language search model based on the knowledge graph.