국내기사
GPT 기반 인공지능을 활용한 국방소요기획 신속화 방안 : OSINT 자동분석과 정량 검증 = A study on the application of GPT-based artificial intelligence for rapid and objective defense requirement planning
This study proposes a method to improve the speed and objectivity of defense requirement planning using GPT-based artificial intelligence. In particular, the study applies a large language model (LLM) to the real-world weapon system, the K9 self-propelled howitzer, analyzing each phase of the requirement planning process-including operational environment analysis, technology evaluation. Open-source defense information (OSINT) from 1989 to 2024 was analyzed using GPT-4 Turbo, which automatically extracted and classified weapon system attributes. The time-series analysis achieved high consistency, with an accuracy of 0.94 and a Macro-F1 score of 0.92. These results demonstrate the applicability of LLMs as data-driven analytical tools throughout all stages of defense requirement planning.