摘要
所谓世界模型,是指能够对物理世界和人类社会进行模拟和理解的人工智能技术图景。Sora、V-JEPA和Genie等新兴人工智能技术,皆以世界模型作为它们的技术原理和发展方向。本文试图从世界模型的整体发展趋势出发,研究其技术机制将如何影响人类社会的隐私机制,进而探讨隐私权以及相关法律制度可能给予的回应。目前人工智能领域存在两种主流世界模型路径,可分别把它们概括为“模拟式世界模型”和“理解式世界模型”。进而指出,世界模型的技术机制可以类比理解为人类社会的四种隐私机制,即:对世界现象的还原,对因果关系的探究,对世界现象的虚构,对社会人格的伪装。并预测随着世界模型的技术发展以及相应社会变迁,现有隐私权制度将进一步被弱化,而个人信息法和数据法将承担起更主要的隐私法律规制功能。因此提出以下建议:国家应该加强对个人信息的行政规制,即在公法层面细化个人信息的类型和行为规范;针对世界模型应用的具体场景和媒介,立法者和监管者应当提供差异化的行为规范和主体责任模式;法律应该针对世界模型隐私问题建立“多方参与治理模式”,鼓励所有社会主体积极参与技术和法律的互动过程。
World models are artificial intelligences that can simulate and understand the world as well as the society.As world models technologies such as Sora,V-JEPA and Genie continue to evolve,social privacy mechanisms and legal privacy institution are affected accordingly.This paper first introduces the two most important theoretical paths of world models.This paper attempts to study how their technological mechanisms will affect the privacy mechanisms of human society from the perspective of the overall development of the world model,and then explore the possible responses to the right to privacy and the related legal system.At present,there are two mainstream world model approaches in the field of artificial intelligence,which can be summarized as“simulation-based world model”and“understanding-based world model”respectively.It is then pointed out that the technical mechanisms of world modeling can be analogously understood as the four privacy mechanisms of human society,i.e.,the reduction of the world phenomenon,the investigation of causality,the fictionalization of the world phenomenon,and the camouflage of the social personality.It is predicted that with the technological development of the world model and the corresponding social changes,the existing privacy regime will be further weakened,and personal information law and data law will assume a more dominant role in the legal regulation of privacy.Therefore,this paper puts forward the following suggestions:the state should strengthen the administrative regulation of personal information,i.e.,refine the types of personal information and behavioral norms at the level of public law;legislators and regulators should provide differentiated behavioral norms and responsibility models for specific scenarios and mediums in which the world model is applied;the law should establish a“multi-party participatory governance model”for world models privacy issues,and encourage the development of a“multi-party participatory governance model”for world models privacy issues.The law should establish a“multi-party governance model”for the privacy issue of world models,and encourage all social actors to actively participate in the interactive process of technology and law.
出处
《地方立法研究》
CSSCI
2024年第3期18-38,共21页
Local Legislation Journal
基金
上海市哲学社会科学规划课题“关键信息基础设施中的隐私问题研究”(2020BFX003)的阶段性成果。
关键词
世界模型
人工智能
隐私机制
个人信息法
数据法
world model
artificial intelligence
privacy mechanisms
personal information law
data law
