生成式人工智能的智能涌现、风险规制与产业调控  

The Emergent Intelligence,Risk Regulation and Industrial Regulation of Generative Artificial Intelligence

作　　者：顾男飞 Gu Nanfei

机构地区：[1]吉林大学法学院

出　　处：《荆楚法学》2023年第3期70-83,共14页Jingchu Law Review

摘　　要：生成式人工智能正逐步实现由弱人工智能向强人工智能的进化。大模型技术的应用使以ChatGPT为代表的生成式人工智能产生智能涌现,出现以自主学习、泛化能力强与交互修正为代表的创新特性,将推动产业升级并驱动第四次工业革命的发展。为针对性回应技术发展的伴生风险,从数据输入、算法运行与内容输出三阶段切入,就关涉的数据治理、算法偏见、著作权侵权及社会治理风险进行类型化讨论,明确法律规制路径。同时为实现风险规制和技术发展的双向互动,在宏观调控视角下,通过明晰数据产权,优化算力供给及激励算法创新等政策供给来促进产业发展,并借助公平竞争审查来维护市场有效竞争。Generative artificial intelligence is gradually evolving from weak to strong AI.The application of comprehensive model technology has enabled generative AI represented by Chat-GPT to exhibit intelligent emergence and innovative characteristics,such as autonomous learning,strong generalization ability,and interactive correction,which will promote industrial upgrading and drive the development of the fourth industrial revolution.To fully respond to the accompanying risks of technological development,the types of data governance,algorithm bias,copyright infringement,and social governance risks involved in the three stages of data input,algorithm operation,and content output are discussed.The legal supervision path is clarified.At the same time,to achieve two-way interaction between risk regulation and technological promotion from a macro-control perspective,it is necessary to combine the three elements of Al:data,computing power,and algorithms through targeted industrial policies such as clarifying data property rights,optimizing computing power supply,and incentivizing algorithm innovation to promote technological development.In addition,fair competition reviews are conducted to maintain market competition.

关 键 词：生成式人工智能 ChatGPT 智能涌现 法律风险 公平竞争审查 

分 类 号：D922.16[政治法律—宪法学与行政法学]