- 期刊
生成式人工智能视域下虚假信息的层级化运作机理与治理
Hierarchical Operation Mechanisms and Governance of Disinformation in the Era of Generative AI
摘要
生成式人工智能在信息生产和传播中的应用促使信息秩序发生变迁，极大地便利了虚假信息的生成与传播。文章聚焦生成式人工智能视域下虚假信息的运行机理，主张对其实施层级化精准治理，即在创设层注重把关，以链式监测、分级管理的数据准入制度，共创新型把关进路，将虚假信息抑制在源头；在撒播层注重研判，消除对话界面所置入的虚假信息落点，并对创设层中浮现的虚假信息进行信息标签的显性呈现与数字水印的隐性嵌入，将虚假信息困囚于对话界面之中，形成扩散阻碍；在流通层注重处置，消除虚假信息的可见性，优化算法对虚假信息的自动化决策过程，借助政府、媒体等公共力量，释放真实透明的媒介信息，形成机动化的虚假信息遮蔽机制，强化信息治理效能。
並列摘要
The widespread integration of generative AI in information production and communication has ushered in a significant paradigm shift in information order, notably expediting the generation and communication of disinformation. This paper delves into the operational mechanisms of disinformation within the realm of generative AI and advocates for a nuanced, hierarchical governance approach. Specifically, at the creation layer, to create a new type of gate-keeping strategy to suppress disinformation at the source of information production, it is necessary to establish a Data Access System with chained monitoring and hierarchical management as its main component. At the dissemination layer, to eliminate the disinformation put in by the user in the dialog interface, trap the disinformation in the dialog interface and form a diffusion obstacle, the disinformation produced in the creation layer is explicitly presented with information labels and implicitly embedded with digital watermarking at the same time. At the circulation layer, the emphasis lies in disposition, obscuring the visibility of disinformation by refining algorithms governing its automated decision-making processes. It also brings into play the power of the government, the media and other public forces to release truthful and transparent information, thereby forming a flexible disinformation filtering mechanism and enhancing the effectiveness of information governance.