基于打击链的智能武器风险分析与控制方法研究

Kill-chain-based analysis of the risks of intelligent weapons and methods for their control

打击链是分析武器和系统等作战要素有效性的一种方法。针对当前智能武器概念缺乏一致定义的问题,本文提出了基于打击链作战方式的智能武器判定方法。针对智能武器认定困难引发的军控措施争论,分析了智能武器两方面的失控风险:人失去对自主武器的控制权与自主武器本身存在不可控因素和缺陷;以及智能武器的问责困境:如何在指挥者、编程者与智能武器之间分配责任。在此基础上,思考了智能武器在未来战争体系对抗中的使用模式,提出了智能武器的风险控制与军备控制方法方面的建议,包括“武器+AI”与“AI+武器”模式探讨、限制智能武器作用域、智能武器应内嵌物理限制系统、限制智能武器的训练数据以及国防采办过程控制等。

Kill chains constitute a good model for analyzing the effectiveness of combat elements such as weapons and systems. Owing to the lack of a uniform definition of the concept of intelligent weapons, this study develops a method for judging intelligent weapons based on the kill-chain framework. In view of the arms control controversies caused by the difficulty in identifying intelligent weapons, two risk factors that could lead to loss of control over intelligent weapons were analyzed: the loss of control over autonomous weapons, and the existence of uncontrollable factors and defects in autonomous weapons. The accountability dilemma of intelligent weapons,that is, the allocation of responsibility among commanders, programmers, and intelligent weapons, was also analyzed. Based on the above, the modes by which intelligent weapons could be deployed in future confrontations were addressed. Proposals for risk management and arms control with respect to intelligent weapons were also provided. The "weapon + AI" and "AI + weapon" deployment modes were examined, and the suggestions for risk management included limiting the scope of intelligent weapons, embedding physical restriction systems in intelligent weapons, limiting training data for intelligent weapons, and controlling the defense acquisition process.