论类ChatGPT通用人工智能治理——基于算法安全审查视角
作者:
作者单位:

(华东交通大学人文社会科学学院,江西 南昌330013)

作者简介:

邹开亮(1976—),男,副教授,主要从事数字法学研究。E-mail:zoukailiang@163.com

通讯作者:

中图分类号:

D90-052

基金项目:

国家社会科学基金一般项目(21BFX126);江西省研究生创新专项资金项目(YC2022-s543)


Research on ChatGPT-like Artificial General Intelligence Governance: From the Perspective of Algorithmic Security Review
Author:
Affiliation:

(College of Humanities and Social Sciences, East China Jiaotong University, Nanchang 330013, China)

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 文章评论
    摘要:

    类ChatGPT具有潜意识形态性,其引入和嵌入应用易招致非传统国家安全威胁。类ChatGPT通用人工智能算法受资本与西方价值观的浸润,冲击国家意识形态安全和情报安全,易诱发意识形态安全风险,并为网络情报的获取和传输提供诸多便利条件;杜撰假舆论,致使社会谣言泛滥,或将挑战社会公共安全;收集存储处理用户数据,严重威胁国家数据安全;阻碍发展中国家崛起的步伐,使产业跨国转移变得缺乏现实意义并重新定义社会人力资源成本上限,或将挟制他国经济发展安全。我国算法审查制度面临专门审查主体缺失、审查范围不明和审查法律规范体系化水平严重滞后等现实问题。应当以总体国家安全观为指引,明确中央科技委员会领导的、国务院数据安全局负责执行的算法安全专门审查主体机制,统筹算法审查宏观设计;充分发挥平台经营者和使用者的协同共治效能,夯实算法审查的社会基础;将算法文化监控、算法可解释性和算法可控性纳入算法审查范围,合理评估算法对国家和社会的冲击,明确算法安全内容;制定“算法安全法”以破解政策敏锐立法滞后的困局,回应算法安全制度的现实需求和总体国家安全的需要,提升国家在科技领域的监管能力,积极应对西方国家算法霸权,加速推进类ChatGPT通用人工智能本土化应用及国产化替代进程。

    Abstract:

    The ChatGPT-like has ideological characteristics, and its introduction and embedded applications are prone to non-traditional national security threats. The ChatGPT-like artificial general intelligence algorithm is fed by capital and western values, which can easily trigger ideological security risks and attack both the country’s ideological security and intelligence security, and provides many convenient conditions for the acquisition and transmission of network intelligence. Fabricating false public opinion and causing widespread social rumors may challenge social public safety, collecting, storing and processing user data poses a serious threat to national data security, and obstructing the pace of a country’s rise may threaten the developing country’s economic development and security, making the transnational transfer of industries impractical and redefining the upper limit of social human resource costs. The algorithm review system in China faces practical problems including the lack of specialized review subjects, unclear review scope, and the inability to implement the review system. Guided by the overall national security concept, we should clarify the specialized review subject of algorithm security led by the Central Science and Technology Commission and implemented by the Data Security Bureau of the State Council to unify the macro design of algorithm review. We should give full play to the synergy between platform operators and users to solidify the social foundation of algorithm review. We should incorporate algorithm culture monitoring, interpretability, and controllability into the scope of algorithm review, reasonably assessing the impact of algorithms on country and society and clarify the content of algorithm security. And we should develop an algorithm security law to ensure the implementation of the algorithm security review system, to enhance the regulatory capacity of the country in the field of technology, actively respond to the algorithm hegemony of western countries, and accelerate the localization application and localization replacement process of ChatGPT-like artificial general intelligence.

    参考文献
    相似文献
    引证文献
引用本文

邹开亮,刘祖兵.论类ChatGPT通用人工智能治理——基于算法安全审查视角[J].河海大学学报(哲学社会科学版),2023,25(6):46-59.(ZOU Kailiang, LIU Zubing. Research on ChatGPT-like Artificial General Intelligence Governance: From the Perspective of Algorithmic Security Review[J]. Journal of Hohai University (Philosophy and Socail Sciences),2023,25(6):46-59.(in Chinese))

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2023-03-25
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2023-12-28
  • 出版日期: