主体弥散化与主体责任的终结:ChatGPT对全球安全实践的影响

Subject Dispersion and the End of Subject Responsibility:Possible Impact of ChatGPT on Global Security Practices

在人工智能时代,以ChatGPT为代表的人工智能通用大模型将可能冲击全球安全互动中的“主体责任制”,造成安全主体的模糊化与责任缺失、安全客体责任的离散化与安全责任追责难等一系列问题。由于人工智能技术由弱向强、迭代演进的技术发展逻辑,不同技术阶段的“主体性危机”与“责任危机”的形态和程度各异。在弱人工智能时代,低智性通用大模型的赋能作用可能会强化安全互动场域中的边缘安全主体责任。安全互动呈现出多重安全主体并存与安全客体责任离散的互动模式。在强人工智能时代,通用大模型被拟造为“类人”主体,成为多重安全主体中的一分子,安全互动呈现出无秩序形态。既有安全主体可能会表现出过度依赖算法和主体权力被侵蚀的特征,而安全客体则表现为普遍卸责和主体责任终结的特点。为解决上述问题,需要从安全主体互动的“主体间性”层面入手,在弱人工智能时代,可利用区块链与人工智能技术形成合力,并细化事前责任;而在强人工智能时代,则可凭借大国协调,在确立智能体“半主体”身份后,建立一种“监护人”制度,在主体间性的逻辑上开展全球安全实践。

In the era of artificial intelligence,the general AI model represented by ChatGPT may impact the “subject responsibility system” in global security interactions,thus giving rise to a series of problems such as the fuzzification and lack of responsibility of security subjects,the discretization of security object responsibility and the difficulties in identifying accountability for security issues.Given that the technological development of artificial intelligence technology is featured by iterative evolution from being weak to being strong,the forms and degrees vary greatly concerning the “subjectivity crisis” and the “responsibility crisis” at different technological stages.In the era of weak artificial intelligence,the empowering effects of the low intelligence general large model may strengthen the responsibility of the marginal security subjects in the field of security interaction,where security interaction presents a mode featuring coexistence of multiple security subjects and the discretization of security object responsibility.In the era of strong artificial intelligence,the general model is designed as a “humanoid” subject,which is a new one among the existing multiple security subjects.The security interaction has taken on irregular forms.The existing security subjects may show the characteristics of over-reliance on algorithm and the erosion of subject power,while security objects may display the general features of shirking their responsibilities and the end of subject responsibility.In order to solve the above problems,we need to start at the “intersubjectivity” level of security subject interaction.In the era of weak artificial intelligence,blockchain and artificial intelligence technology can be utilized to form synergy and refine prior responsibilities.In the era of strong artificial intelligence,it is advisable to forge a “guardian” system based on coordination among major countries and the establishment of the “semi-subject” identity of intelligent agents so that global security practices can be carried out along the logic line of intersubjectivity.