人工智能伦理问题与安全风险治理的全球比较与中国实践Governance of Ethical Challenges and Safety Risks of Artificial Intelligence: Global Comparisons and Practice in China
贾开,薛澜
摘要(Abstract):
从全球范围来看,利益相关体已经提出了诸多人工智能治理原则以试图通过"软体系"的方式应对人工智能发展和应用所引发的伦理问题与安全风险。本文指出,人工智能的主体性挑战、代码作为规则重要性的提升以及"黑箱性"的存在构成了以"软体系"应对人工智能伦理问题与安全风险的特殊原因,而推进理念共识、聚焦局部议题、创新治理机制则构成了其具体内涵。虽然现有方案有利于推动人工智能伦理问题与安全风险的治理进程,但其在关键概念上的模糊、具体内容上的缺失或争议以及决策过程的不足,也引导人们产生更多反思。中国近年来在三个方向积极推进并于2021年初形成了《网络安全标准实践指南——人工智能伦理安全风险防范指引》的阶段性成果,该文件界定了人工智能治理的相关主体范围,总结了人工智能伦理问题与安全风险的具体类型,并明确了相关主体所应履行的治理职责。结合全球比较和中国实践的分析,文章进一步对未来人工智能伦理问题与安全风险治理的发展可能提出了三点展望。
关键词(KeyWords): 人工智能;伦理问题;安全风险;治理反思
基金项目(Foundation):
作者(Author): 贾开,薛澜
参考文献(References):
- 贾开.2019.人工智能与算法治理研究[J].中国行政管理,(1):17-22.Jia K.2019.Artificial intelligence and algorithm governance research[J].Chinese Public Administration,(1):17-22.(in Chinese)
- 沈伟伟.2019.算法透明原则的迷思---算法规制理论的批判[J].环球法律评论,(6):20-39.Shen W W.2019.The myth of the algorithmic transparency principle:A critique of the algorithm regulation theory[J].Global Law Review,(6):20-39.(in Chinese)
- 吴文俊.2019.吴文俊全集:数学思想卷[M].北京:科学出版社:66.Wu W J.2019.The complete works of Wu Wen-tsun:Mathematical thoughts[M].Beijing:Science Press:66.(in Chinese)
- 薛澜,赵静.2019.走向敏捷治理:新兴产业发展与监管模式探究[J].中国行政管理,(8):28-34.Xue L,Zhao J.2019.Toward agile governance:The pattern of emerging industry development and regulation[J].Chinese Public Administration,(8):28-34.(in Chinese)
- Acemoglu D,Restrepo P.2020.The wrong kind of AI?Artificial intelligence and the future of labour demand[J].Cambridge Journal of Regions,Economy and Society,13(1):25-35.
- Baron B,Musolesi M.2020.Interpretable machine learning for privacy-preserving pervasive systems[J].IEEE Pervasive Computing,19(1):73-82.
- Benkler Y.2019.Don’t let industry write the rules for AI[J].Nature,569(7754):161.
- Bogost I.2012.The New Aesthetic Needs to Get Weirder[J].The Atlantic,13(4):1-8.
- Corbett-Davies S,Pierson E,Feller A,et al.2017.Algorithmic decision making and the cost of fairness[C]//Proceedings of the 23rd ACM Sigkdd International Conference on Knowledge Discovery and Data Mining.Halifax:ACM,797-806.
- Future of Life Institute.2017.Asilomar AI principles[EB/OL].https://futureoflife.org/aiprinciples/.
- Gebru T,Morgenstern J,Vecchione B,et al.2018.Datasheets for datasets.ar Xiv:1803.09010.
- Greene D,Hoffmann A L,Stark L.2019.Better,nicer,clearer,fairer:A critical assessment of the movement for ethical artificial intelligence and machine learning[C]//Proceedings of the 52nd Hawaii International Conference on System Sciences.Grand Wailea:ScholarSpace,1-10.
- Hagendorff T.2020.The ethics of AI ethics:An evaluation of guidelines[J].Minds and Machines,30(1):99-120.
- Lobel O.2012.New governance as regulatory governance[M]//Levi-Faur D.The Oxford Handbook of Governance.Oxford:Oxford University Press.
- McNamara A,Smith J,Murphy-Hill E.2018.Does ACM’s code of ethics change ethical decision making in software development?[C]//Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering.Lake Buena Vista:ACM,729-733.
- Mittelstadt B D,Russell C,Wachter S.2019.Explaining explanations in AI[C]//Proceedings of Conference on Fairness,Accountability,and Transparency.Atlanta:ACM,279-288.
- Morozov E.2019.Digital socialism,The calculation debate in the age of big data[J].New Left Review,116/117:33-67.
- OECD.2019.OECD principles on AI[EB/OL].https://www.oecd.org/going-digital/ai/principles/.
- Polanyi M.2009.The Tacit Dimension[M].Chicago:University of Chicago Press:x.
- Sabel C F,Zeitlin J.2012.Experimentalist governance[M]//Levi-Faur D.The Oxford Handbook of Governance.Oxford:Oxford University Press.
- Sweeney L.2013.Discrimination in online ad delivery:Google ads,black names and white names,racial discrimination,and click advertising[J].Queue,11(3):10-29.
- UK House of Lords.2017.AI in the UK:Ready,willing and able?[EB/OL].https://publications.parliament.uk/pa/ld201719/ldselect/ldai/100/10002.htm.
- Watson B C.2011.Barcode empires:Politics,digital technology,and comparative retail firm strategies[J].Journal of Industry,Competition and Trade,11(3):309-324.
- Wirtz B W,Müller W M.2019.An integrated artificial intelligence framework for public management[J].Public Management Review,21(7):1076-1100.
- (1)以《新一代人工智能发展规划》《国家新一代人工智能标准体系建设指南》等文件为代表。
- (2)以《新一代人工智能治理原则——发展负责任的人工智能》《人工智能北京共识》等文件为代表。
- (3)包括《信息安全信息安全技术虹膜识别系统技术要求》《信息安全技术基于可信环境的生物特征识别身份鉴别协议框架》《信息安全技术指纹识别系统技术要求》《信息安全技术汽车电子系统网络安全指南》《信息安全技术车载网络设备信息安全技术要求》《信息安全技术智能家居安全通用技术要求》《信息安全技术智能门锁安全技术要求和测试评价方法》等。
- (1)本文所指治理“软体系”是区别于具有明确规则和约束力、强制力的“硬法体系”而言的,其旨在分析治理风险、提出治理目标、构建治理机制、提出治理方案,可被视为“硬法体系”形成之前的治理共识的达成过程,或者因难以具体、明确地界定不同主体的责任、权利、义务而采取的其他治理进程。
- (1)例如人工智能研究领域到目前为止也未能对“人工智能”的定义形成共识,主要的分歧之一在于部分研究者认为只要在结果层面重复人类行为即可称为“人工智能”,但其他研究者认为要在过程中也模仿人类的思维模式才能被视为“智能”。
- (1)我国企业的代表性参与行为包括腾讯发布《智能时代的技术伦理观——重塑数字社会的信任》,百度参与Partnership on AI等。