An Empirical Study on the Development of Risk Assessment Indicators for Defense Artificial Intelligence Weapon Systems : Compound Focusing on Field Operators Below Company Level

Le résumé fourni par la source

The advancement of artificial intelligence technology and unmanned weapon systems such as drones has led to active development of AI-based weapon systems in the defense sector. While addressing numerous operational demands, AI weapon systems may cause harm due to malfunctions or misrecognitions. However, research to prevent such incidents remains insufficient. This study proposes risk assessment indicators to ensure the safe operation of defense AI weapon systems, focusing primarily on operators at the company level and below. To develop these indicators, a literature review and case analysis were conducted, followed by creating a survey questionnaire. The survey items were designed to be universal by incorporating joint functionalities to distinguish “risk areas,” at the same time functions to be implemented were classified as “items,” and causes inhibiting functionality implementation were proposed as “measurement indicators.” Two rounds of Delphi surveys were conducted with 30 experts specializing in the defense sector, artificial intelligence, and weapon systems. Statistical analysis verified the reliability and validity of the proposed indicators, categorizing risks into six areas with 19 items, ultimately deriving 65 evaluation indicators. These indicators are expected to contribute to enhancing the operational safety of defense AI weapon systems. Furthermore, this study provides a foundation for developing more sophisticated risk assessment indicators tailored to individual weapon systems.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.