图书情报工作 ›› 2021, Vol. 65 ›› Issue (12): 47-56.DOI: 10.13266/j.issn.0252-3116.2021.12.004

• 研究论文 • 上一篇    下一篇

智能情报分析中算法风险及其规制研究

张涛1, 马海群2   

  1. 1. 黑龙江大学信息管理学院 哈尔滨 150080;
    2. 黑龙江大学信息资源管理研究中心 哈尔滨 150080
  • 收稿日期:2020-11-22 修回日期:2021-02-25 出版日期:2021-06-20 发布日期:2021-07-03
  • 作者简介:张涛(ORCID:0000-0002-3367-4541)副教授,博士研究生,硕士生导师,E-mail:zhangtao@hlju.edu.cn;马海群(ORCID:0000-0002-2091-7620)教授,博士,博士生导师。
  • 基金资助:
    本文系国家社会科学基金重点项目"总体国家安全观下的国家情报工作制度创新研究"(项目编号:20ATQ004)研究成果之一。

Research on Algorithm Risk and Regulation in Intelligent Intelligence Analysis

Zhang Tao1, Ma Haiqun2   

  1. 1. School of Information Management, Heilongjiang University, Harbin 150080;
    2. Research Center of Information Resource Management, Heilongjiang University, Harbin 150080
  • Received:2020-11-22 Revised:2021-02-25 Online:2021-06-20 Published:2021-07-03

摘要: [目的/意义] 近年来,人工智能给国家情报工作带来思维理念和方法技术的转变,使得智能情报分析逐渐成为服务国家情报事业创新发展的重要工作之一。对智能情报分析中的算法风险及其规制进行研究,可以避免人工智能算法所带来的安全风险,减少限制情报工作发展风险方面的因素。[方法/过程] 基于对智能情报分析核心算法及研究问题的阐释,结合实际应用场景对算法风险形成原因、产生后果及算法风险各因素间的交互关系进行分析。通过事前评估、事中监管、事后问责的手段建立循序渐进的智能情报分析算法规制框架。[结果/结论] 分析"情报"茧房的形成机理,针对智能情报分析中所存在的算法黑箱、算法缺陷、算法操控等风险,提出事前、事中、事后的算法规制良性循环与协调发展建议,并认为正视算法的两面性也是防范与化解算法风险的有效途径。

关键词: 智能情报分析, 算法风险, 算法规制, 智能算法, “情报”茧房

Abstract: [Purpose/significance] In recent years, artificial intelligence has brought changes in thinking, concepts, methods and techniques to national intelligence work, making intelligent intelligence analysis gradually become one of the important tasks for the innovation and development of national intelligence. This paper studies the algorithm risk and its regulation in intelligent intelligence analysis to avoid the security risks caused by artificial intelligence algorithms and reduce the factors that restrict the development of intelligence work.[Method/process] Based on the interpretation of the core algorithms and research issues of intelligent intelligence analysis, combined with actual application scenarios, the causes and consequences of algorithmic risk and the interaction between the factors of algorithmic risk are analyzed. Establish a gradual regulatory framework for intelligent intelligence analysis algorithms through pre-assessment, mid-event supervision, and post-event accountability.[Result/conclusion] This article analyzes the formation mechanism of the "intelligence" cocoon room, and proposes a virtuous cycle of algorithm regulation and coordinated development before, during, and after the risks of algorithm black boxes, algorithm defects, and algorithm manipulation in intelligent intelligence analysis, and believes that Facing the two sides of the algorithm is also an effective way to prevent and defuse the risk of the algorithm.

Key words: intelligent intelligence analysis, algorithmic risk, algorithmic regulation, intelligent algorithm, "intelligence" cocoon room

中图分类号: