摘要
[目的/意义]生成式人工智能(Generative AI)的突破性进展带来了人工智能生成内容(AIGC)的爆炸式增长,不可避免地将人们置于信息过载、信息噪声、信息安全等的负面影响之下,使得社会信息治理面临新的挑战。分析和探讨现有AIGC事故的特征属性,对我国生成式人工智能治理有参考借鉴作用。[研究设计/方法]基于AI事故数据库(AIID),以AIGC相关事故报道为样本进行内容分析,探析现有AIGC事故的类型、原因、损害对象和应对措施。[结论/发现]AIGC事故影响客体的多元性、波及范围的广泛性、潜在危害的复杂未知性,导致任何单一行动主体的资源和能力都无法有效应对危机,需要政府、企业、社会三方行动主体形成“多元+协调+制衡”的治理参与模式,并在“情境-意识-行动”的行动框架下开展信息治理。[创新/价值]引入了AIID作为案例来源数据库,提供了关于现有AIGC事故相关细节的直观论证,并通过内容分析形成了AIGC事故分析三级类目框架。构建的生成式人工智能治理行动框架有助于从宏观视角促进我国生成式人工智能治理的探索和实践。
[Purpose/Significance]The breakthrough of Generative Artificial Intelligence(Generative AI)has led to the explosive growth of Artificial Intelligence Generated Content(AIGC),which inevitably cause people to be negatively affected by information overload,information noise,information security,and other related issues,making social information governance face new challenges.This paper aims to analyze and discuss the characteristic attributes of AIGC incidents,so as to provide a reference for Generative AI governance in China.[Design/Methodology]Based on AI Incident Database(AIID)and taking AIGC-related incident reports as samples for content analysis,the types,causes,damage objects and countermeasures of existing AIGC incidents were discussed.[Findings/Conclusion]The diversity of the objects affected by AIGC incidents,the wide distribution of the scope,and the complex unknown of the potential harm,result in the resources and capabilities of any single actor not being able to effectively deal with the crisis.It is necessary for actors of government,enterprises and society to form a governance participation model of"diversity+coordination+checks and balances",and to carry out information governance under the action framework of"context-consciousness-action".[Originality/Value]This paper introduces AIID as a case source database,provides an intuitive demonstration of the relevant details of existing AIGC incidents and forms a three-level category framework for AIGC incident analysis through content analysis.The action frame of Generative AI governance formed in this study is helpful to promote the exploration and practice of Generative AI governance from a macro perspective.
作者
朱禹
陈关泽
陆泳溶
樊伟
ZHU Yu;CHEN Guanze;LU Yongrong;FAN Wei(School of Information Management,Nanjing University,Nanjing,210023;Department of Computer Science and Engineering,The Chinese University of Hong Kong,Hong Kong,999077;Pittsburgh Institute,Sichuan University,Chengdu,610207;Sichuan University Library,Chengdu,610065)
出处
《图书情报知识》
CSSCI
北大核心
2023年第4期41-51,共11页
Documentation,Information & Knowledge
关键词
生成式人工智能
人工智能生成内容
信息治理
行动框架
内容分析
Generative artificial intelligence
Artificial Intelligence Generated Content(AIGC)
Information governance
Action framework
Content analysis
作者简介
通讯作者:朱禹(ORCID:0000-0002-2548-828X),硕士研究生,研究方向:信息资源管理,Email:13350057427@163.com;陈关泽(ORCID:0009-0008-6974-4864),硕士研究生,研究方向:信息资源管理､可解释人工智能,Email:1625205886@qq.com;陆泳溶(ORCID:0009-0004-7629-3281),本科生,研究方向:人工智能,Email:rita111585@icloud.com;樊伟(ORCID:0009-0002-6278-910X),硕士,馆员,研究方向:信息资源管理,Email:535723984@qq.com｡。
