论“人工智能法”框架体系的多维动态模式  

On The Multidimensional Dynamic Model of the "Artificial Intelligence Law" Framework

作　　者：马国洋 Ma Guoyang

机构地区：[1]北京交通大学法学院

出　　处：《东方法学》2025年第1期186-200,共15页Oriental Law

基　　金：国家社会科学基金重点项目“大数据侦查的程序控制与证据适用研究”(项目批准号:19AZD024);中央高校基本科研业务费专项资金“生成式人工智能在事实认定中的应用方式与风险规制研究”(项目批准号:2023JBW005)的阶段性研究成果。

摘　　要：“人工智能法”框架体系的多维动态模式是顺应人工智能治理发展趋势的产物,其可以有效满足人工智能治理精细化的需求,同时也符合立法的内在要求和抢占话语权的需要。多维动态模式是三个基本维度的有机组合,分别是基于主体进行治理的价值链维度、基于流程进行治理的生命周期维度以及基于不同场景进行治理的场景维度。在组合方式上,多维动态模式要求以人工智能的生命周期为主线,在不同环节为价值链中的不同主体赋予相应的权利和责任方式,并且根据特殊场景在各个环节设置针对性规则。其动态化调整则应基于敏捷治理的原理,通过设立“人工智能法”框架体系的实时监测、多元合作和信息反馈机制加以实现。这种多维动态模式有效结合了人工智能治理的各单一模式,可以弥补单一模式的缺陷,并更好地适应技术发展。The multidimensional dynamic model of the Artificial Intelligence Law framework is a product of the evolving trends in artificial intelligence governance, which can effectively meet the needs of the refinement of artificial intelligence governance, and at the same time meets the intrinsic requirements of legislation and the need to seize the right to speak. The multidimensional dynamic model is an organic combination of three basic dimensions: governance based on actors(value chain dimension), governance based on processes(lifecycle dimension), and governance based on different contexts(scenario dimension). In terms of composition, the multidimensional dynamic model requires that the life cycle of AI is the main line, assigning corresponding rights and responsibilities to different stakeholders in the value chain at various stages, and designing targeted rules for each stage based on specific scenarios. Its dynamic adjustment is based on agile governance principles, achieved through real-time monitoring, multifaceted collaboration,and information feedback mechanisms within the Artificial Intelligence Law framework. This multidimensional dynamic model effectively combines various single-mode approaches, addressing the limitations of single models and better adapting to technological development.

关 键 词：人工智能 人工智能法 敏捷治理 框架体系 多维动态模式 人工智能治理 

分 类 号：DF0[政治法律—法学理论]