人工智能武器犯罪类型界分与风险控制

Classification and Risk Control of Artificial Intelligence Weapon Crime

随着人工智能技术不断发展,武器系统已成为人工智能研发应用新领域。当前人工智能武器的发展路径呈现二元化趋势:“武器+AI”与“AI+武器”。从大数据算法到深度学习,多类型化的人工智能武器犯罪问题也使“科技改变战争”这句话有了另一面向的隐喻。以人工智能武器“主动性”为标准,划定人工智能武器犯罪类型是应对安全风险的核心问题。从挑战国际法规则、改变战争核心逻辑、异化交战权属性等方面,研判人工智能武器犯罪风险,是事关时代的理论演进。基于安全价值认知,面向源头治理,从立法规制、技术标准、监护制度等预防性、综合性策略入手,积极构建针对已然和未然犯罪的防范体系,渐进形成人与人工智能在“行为控制”与“决策能力”上的和谐关系。

With the continuous development of artificial intelligence technology,weapon system in military field has become a new territory of artificial intelligence.At present,the development path of artificial intelligence weapons presents a dual trend,either“weapons+AI”or“AI+weapons”.From big data algorithm to deep learning,the problem of multiple types of artificial intelligence weapon crime also gives the phrase“technology changing war”another metaphor.The core issues to deal with security risks are to take the“initiative”of artificial intelligence weapons as the standard and define the types of artificial intelligence weapon crimes.It is a challenging theoretical approach to study and judge the criminal risk of artificial intelligence weapons from the aspects of challenging the rules of international law,changing the core logic of war and dissimilating the attribute of belligerency.Based on logical cognition and source control and starting with preventive and comprehensive strategies such as legislative regulation,technical standards and guardianship system,this paper constructs a prevention system for existing and pending crimes,so as to gradually form a harmonious relationship between human and artificial intelligence in“behavior control”and“decision ability”.