Risk Perception and Acceptance of Autonomous Weapon Systems: A Comparison Between Public and Military Personnel

Abstract

This study aims to examine the acceptance of autonomous weapon systems (AWS) among public and military personnel, distinguishing between acceptance of development and adoption and acceptance of deployment and use, and further disaggregating use acceptance by mission context. Based on just war theory, particularly jus in bello principles, this study conceptualizes AWS acceptance not as a technological preference but as a normative evaluation of whether AWS are perceived to satisfy core requirements in the conduct of war, such as distinction and proportionality. For this purpose, a survey was conducted with a nationally stratified sample of public (N = 1,222) and a sample of active-duty military personnel (N = 155). Regression analyses were conducted to compare the effects of demographic characteristics, value orientations, perceptions related to just war principles and risk, and attitudes toward artificial intelligence on AWS acceptance across stages and contexts. The results show that patterns of acceptance differ systematically between public and military personnel. Military respondents exhibit significantly higher acceptance of AWS development and adoption than public, whereas they show lower acceptance of AWS deployment and use. The results also show that perceptions related to just war principles and risk are the strongest predictors of acceptance in both groups, and their effects are particularly pronounced for use-related judgments. In contrast, demographic factors and general AI attitudes play relatively secondary roles. Notably, higher objective AI knowledge is associated with lower acceptance of AWS use. These findings suggest that AWS acceptance is structured by roles, normative perceptions, and risk-related evaluations rather than by technological optimism alone. Furthermore, the lower use acceptance among military personnel may reflect organizational and contextual factors that are not fully captured in survey responses, indicating the need for follow-up qualitative analysis. We conclude governance debates on autonomous weapon systems should move beyond treating social acceptance of AWS as a uniform stance, and instead adopt a differentiated approach that recognizes how judgment logics vary according to the object of acceptance and the contextual conditions under which evaluations are formed.