인공지능을 통한 전쟁수행은 정당한가 - 자율무기체계를 통한 국제인도법 준수가능성을 중심으로 -

This article examines the question whether an armed conflict through the Artificial Intelligence(AI) is justified. For this purpose, it explores technical characteristics of the AI, the meaning of autonomy in weapon systems and (in)compatibility of principles of international humanitarian law to a warfare performed by Autonomous Weapon Systems(AWS). AWS denotes a type of weapons which is capable of choosing and engaging military targets without additional involvement by human soldiers, once being operated. In this regard, the current criticism concerning the utilization of the AI in complex decision-making, such as fairness, ethics and explicability is also related to the AWS. In order to relieve these ethical and legal concerns on the AWS, it is required to maintain the fundamental human control and supervision in the course of selecting and targeting the military targets when using the AWS. This article considers whether the employment of the AWS in the military collides with the current rules of international humanitarian law, and if the current principles of international humanitarian law do not adequately regulate the AWS what alternative options can be suggested. Concerning the employment of the AWS, it seems very difficult for state parties to comply with the principle of distinction and the principle of proportionality. It is partly due to the changed environment of today’s military conflict which is performed in situations where distinction between civilians and combatants is difficult. Furthermore, it is because proportionality principle requires state parties to duly compare estimated civilian damages with estimated military interests before making a certain strategic decision. It is not considered as strict arithmetic examination but as delicate value judgment. Considering this, it is impossible to comply with principles of humanitarian law only through the AWS at the current AI technology level. The AWS is not prohibited as such in current international law, but states and relevant interested parties shall come up with alternative regulations on the AWS in order to prevent the possible violation of international law when employing the AWS in an international armed conflict. States shall make best efforts to restrict the harm involving the AWS, in a way to make a binding international agreement and to establish the meaningful human control principle.