科技智囊Issue(11)：42-52,11.DOI:10.19881/j.cnki.1006-3676.2023.11.06
生成式人工智能技术的安全风险及其防范策略
Security Risks of Generative Artificial Intelligence Technology and Its Prevention Strategies
摘要
Abstract
In the era of big data,the development of generative artificial intelligence represented by ChatGPT contains opportunities for technological revolution and may also bring potential social risks.In order to encourage the innovative development of generative artificial intelligence technology and avoid the security risks it brings,it is necessary to actively propose and construct security risk prevention strategies for GAI technology.The article takes contemporary emerging generative artificial intelligence technology as the research object,follows the paradigm of systematic literature review,explains the connotation and operating mechanism of generative artificial intelligence technology,and further proposes that the security risks of generative artificial intelligence technology can be summarized as ideological risks,social security risks,government construction and governance risks,and political security risks.This type of classification not only confirms the comprehensive social impact of the technology,but also reveals the internal mechanism of risk generation.On this basis,the reasons for the security risks of generative artificial intelligence technology were examined,mainly including:algorithm reinforcement of human bias,the rise of algorithm hegemony,the inversion of human-machine relationships,and human self questioning.In order to build a responsible and forward-looking generative artificial intelligence technology security risk prevention strategy,we should adhere to strengthening the leadership of the Party,strengthening government responsibility,promoting social coordination,and enhancing legal protection to achieve long-term governance of technology security risks,in order to provide reference for serving the modernization of the national governance system and capacity.关键词
ChatGPT/生成式人工智能/预训练模型/安全风险/防范策略Key words
ChatGPT/Generative artificial intelligence/Pre-trained model/Safety risks/Preventive strategies分类
社会科学引用本文复制引用
郭佳楠,陈婉莹..生成式人工智能技术的安全风险及其防范策略[J].科技智囊,2023,(11):42-52,11.基金项目
本文系国家社科基金重点项目"大数据背景下的集体记忆与政治认同研究"(项目编号:22AKS013)、2023年度河南省社科联调研课题"当代新兴ChatGPT技术的伦理风险及其治理路径研究"(项目编号:SKL-2023-252)、郑州工商学院科研创新项目"基于ChatGPT类生成式人工智能的红色文化传播现状与策略研究"(项目编号:2023-KYZD-01)、2023年度郑州工商学院校级教育教学改革研究重点项目"中国特色工匠精神视域下民办高校大学生创新能力培养策略研究"(项目编号:GSJG2023003)研究成果. （项目编号:22AKS013）
