生成式人工智能对信息治理的挑战与应对  

The Challenges and Responses of Generative Artificial Intelligence to Information Governance

作　　者：林雨佳 LIN Yu-jia(Law School,Shanghai University,Shanghai 200444,China)

机构地区：[1]上海大学法学院,上海200444

出　　处：《苏州大学学报(哲学社会科学版)》2025年第2期104-115,共12页Journal of Soochow University(Philosophy & Social Science Edition)

基　　金：国家社会科学基金项目“数字化时代网络犯罪罪量认定研究”(项目编号:24CFX093)的阶段性成果。

摘　　要：生成式人工智能能够“生成”新的信息,这一技术特点导致其在获取、处理和输出信息的过程中,存在不同于既往人工智能的信息治理隐患。在生成式人工智能运行机制下,信息并非单向流动。无监督的预训练和用户的参与意味着存在新的影响“生成”信息的因素,研发者对信息的控制力下降。信息非单向流动容易导致归责路径不清,研发者对信息的控制力下降导致不同主体之间的责任分配存在困难。虽然信息生成是连续的,但生成式人工智能的信息治理宜采用隔离式的应对方案:隔离信息生成的不同过程、隔离不同责任主体、隔离技术治理和规范治理以及隔离通用技术属性和专项服务属性。隔离式应对方案分别从过程、主体、性质、层级四个方面提升信息治理效用,有效解决信息非单向流动和研发者对信息控制力下降带来的信息治理新问题。Generative Artificial Intelligence can“generate”new information,which leads to the hidden dangers of information governance that are different from those in the past in the process of acquiring,processing and outputting information.In generative AI,information does not flow in one direction.Unsupervised pre-training and user involvement mean that there are new factors influencing the“generation”of information,and the developer has less and less control over the information.The non-unidirectional flow of information tends to make the path of responsibility unclear,and the developer’s declining control over information leads to the difficulty of assigning responsibility among different subjects.Although the information generation is continuous,the information governance of generative AI should adopt an isolated solution:separate different processes of information generation,separate different responsibility subjects,separate technology governance and normative governance,and separate general technology attributes and special service attributes.The isolated response plan improves the effectiveness of information governance from four aspects:process,subject,nature,and level,respectively,and effectively solves the new problems of information governance caused by the non-unidirectional flow of information and the declining control of information by developers.

关 键 词：生成式人工智能 信息治理 技术治理 责任主体 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] D99[自动化与计算机技术—控制科学与工程]