| 1,982 | 22 | 527 |
| 下载次数 | 被引频次 | 阅读次数 |
教育人工智能伦理研究的“原则方法”旨在通过伦理原则的构建与教授消解教育领域应用人工智能技术出现的伦理风险。然而，当前教育人工智能伦理原则存在有限性，制约了其对教育实践的影响力。就自身而言，教育人工智能伦理原则在构建过程中忽略了其在不同行业领域、不同教育应用甚至同一框架内的个性特征或关系。就实践而言，教育人工智能伦理原则既无法提升目标对象将伦理原则付诸实践的意愿，也缺乏强制其遵守这些价值观和规范的机制。教育人工智能伦理原则有限性的消解需要在完善原则自身的同时引入美德与法律，进而构建“原则-美德-法律”框架。这一框架的构建要义是使原则、美德、法律三大层次针对且有效地应对伦理原则有限性的各个方面，做到确定可靠切实的伦理原则、开展明确可行的美德培育、构筑稳健有效的法律保障。
Abstract：The “principled approach” to ethical research in educational artificial intelligence aims to construct and teach ethical principles to mitigate the ethical risks associated with the application of AI technology in education. However, current ethical principles for educational AI are limited, which constrains their impact on educational practices. Specifically, in constructing these principles, individual characteristics or relationships within different industry sectors, educational applications, and even within the same framework have been overlooked. Moreover,these principles themselves lack the ability to influence the willingness of the target objects to put them into practice, and there is a lack of mechanisms that can effectively enforce compliance with these values and norms. Ethics, virtues, and laws together constitute the ethical foundation of artificial intelligence. To overcome the limitations of ethical principles for educational AI, it is necessary to introduce virtues and laws and construct a “principle-virtue-law” framework. The key to this framework is to address all aspects of the limitations of ethical principles by developing reliable and viable ethical principles, conducting clear and feasible virtue cultivation and building robust and effective legal safeguards.
[1]Seldon A,Abidoye O.The fourth education revolution[M].London:Legend Press Ltd,2018.
[2]Hagendorff T.AI ethics and its pitfalls:not living up to its own standards?[J].AI and Ethics,2023,3(3):329-336.
[3][4]白钧溢,于伟.超越“共识”：教育人工智能伦理原则构建的发展方向[J].中国电化教育,2023,(6):9-17+24.
[5][6]Adams C,Pente P,et al.Artificial Intelligence ethics guidelines for K-12education:A review of the global landscape[C].Cham:Springer,2021.24-28.
[7]Tuomi I.A framework for socio-developmental ethics in educational AI[C].Hawaii:University of Hawaii,2023.6208-6217.
[8]古天龙.人工智能伦理及其课程教学[J].中国大学教学,2022,(11):35-40.
[9]Morley J,Floridi L,et al.From what to how:an initial review of publicly available AI ethics tools,methods and research to translate principles into practices[J].Science and engineering ethics,2020,26(4):2141-2168.
[10]Kempt H,Freyer N,et al.Justice and the Normative Standards of Explain ability in Healthcare[J].Philosophy&Technology,2022,35(4):248-266.
[11]IBM Corporation.Everyday ethics for artificial intelligence[EB/OL].https://www.ibm.com/downloads/cas/VDO5W3JK?mhsrc=ibmsearch_a&mhq=Everyday%20ethics%20for%20artificial%20intelligence,2019-05-07.
[12][14]Rességuier A,Rodrigues R F.The Ethics of Artificial Intelligence in Education:Practices,Challenges,and Debates[M].London:Taylor&Francis,2022.
[13]Holmes W,Tuomi I.State of the art and practice in AI in education[J].European Journal of Education,2022,57(4):542-570.
[15][32]Whittlestone J,Nyrup R,et al.The role and limits of principles in AIethics:towards a focus on tensions[C].New York:ACM,2019.195-200.
[16]Mc Namara A,Smith J,et al.Does ACM’s code of ethics change ethical decision making in software development?[C].New York:ACMPress,2018.729-733.
[17]Munn L.The uselessness of AI ethics[J].AI and Ethics,2022,1(1):1-9.
[18]Vakkuri V,Kemell K K,et al.Ethically Aligned Design of Autonomous Systems:Industry Viewpoint and an Empirical Study[J].Electronic Journal of Business Ethics and Organization Studies,2022,27(1):4-15.
[19]Vakkuri V,Kemell K K,et al.“This is just a prototype”:How ethics are ignored in software startup-like environments[C].Copenhagen:Springer International Publishing,2020.195-210.
[20][25][29]Rességuier A,Rodrigues R.AI ethics should not remain toothless!A call to bring back the teeth of ethics[J].Big Data&Society,2020,7(2):1-5.
[21]Campolo A,Sanfilippo M,et al.AI now 2017 report[EB/OL].https://assets.ctfassets.net/8wprhhvnpfc0/1A9c3ZTCZa2KEYM64Wsc2a/8636557c5fb14f2b74b2be64c3ce0c78/_AI_Now_Institute_2017_Report_.pdf,2018-10-02.
[22]Knight,W.Google appoints an‘AI Council’to head off controversy,but it proves controversial[EB/OL].https://www.technologyreview.com/2019/03/26/136376/google-appoints-an-ai-council-to-head-offcontroversy-but-it-proves-controversial/,2019-03-26.
[23]Koniakou V.From the“rush to ethics”to the“race for governance”in Artificial Intelligence[J].Information Systems Frontiers,2023,25(1):71-102.
[24]吴红,杜严勇.人工智能伦理治理:从原则到行动[J].自然辩证法研究,2021,37(4):49-54.
[26][36][37][38]Hagendorff T.A virtue-based framework to support putting AIethics into practice[J].Philosophy&Technology,2022,35(3):1-24.
[27]Floridi L.Soft ethics and the governance of the digital[J].Philosophy&Technology,2018,31(1):1-8.
[28]Grosz B J,Grant D G,et al.Embedded Ethi CS:integrating ethics across CSeducation[J].Communications of the ACM,2019,62(8):54-61.
[30]Anscombe G E M.Modern moral philosophy[J].Philosophy,1958,33(124):1-19.
[31]Nemitz P.Constitutional democracy and technology in the age of artificial intelligence[J].Philosophical Transactions of the Royal Society A,2018,2133(376):1-10.
[33]张济洲,黄书光.美德是否可教--论苏格拉底的德性教化[J].教育研究,2013,34(4):76-81.
[34]Harris Jr C E.The Good Engineer:Giving Virtue Its Due in Engineering Ethics[J].Science and Engineering Ethics,2008,14(2):153-164.
[35]Vallor S.Technology and the virtues:A philosophical guide to a future worth wanting[M].Oxford:Oxford University Press,2016.
[39]Bezuidenhout L,Ratti E.What does it mean to embed ethics in data science?An integrative approach based on microethics and virtues[J].AI&Society,2021,36(3):939-953.
[40]Khan A A,Akbar M A,et al.AI Ethics:An Empirical Study on the Views of Practitioners and Lawmakers[J].IEEE Transactions on Computational Social Systems,2023,12(3):1-23.
[41]Papyshev G,Yarime M.The state’s role in governing artificial intelligenc e:development,control,and promotion through national strategies[J].Policy Design and Practice,2023,6(1):1-24.
[42]Erdélyi O J,Goldsmith J.Regulating artificial intelligence:Proposal for a global solution[C].New York:ACM,2018.95-101.
[43]Ebers M,Hoch V R S,et al.The European commission’s proposal for an artificial intelligence act-a critical assessment by members of the robotics and ai law society[J].Multidisciplinary Scientific Journal,2021,4(4):589-603.
[44]de Almeida P G R,dos Santos C D,et al.Artificial intelligence regulation:a framework for governance[J].Ethics and Information Technology,2021,23(3):505-525.
[45][47]Stahl B C,Rodrigues R,et al.A European Agency for Artificial Intelligence:Protecting fundamental rights and ethical values[J].Computer Law&Security Review,2022,45(1):1-25.
[46]白钧溢.教育人工智能伦理治理:现实挑战与实现路径[EB/OL].http://kns.cnki.net/kcms/detail/50.1028.G4.20231211.1217.002.html,2024-01-26.
(1)“原则方法”这一概念由艾伦·图灵研究所布伦特·米特尔斯塔德(Brent Mittelstadt)教授于2019年在分析国际人工智能伦理研究现状后提出，这一概念目前已得到广泛认可并被诸多研究使用。参见Mittelstadt B.Principles alone cannot guarantee ethical AI[J].Nature machine intelligence,2019,1(11):501-507。
(1)这类研究主要是在宏观伦理框架的基础上增加符合本学段特征的伦理原则，如在K-12阶段侧重强调包括以学生为中心、差异化和个性化学习等方面的教学适当性，而在高等教育阶段，关注更多的则是人工智能素养。
(1)“伦理漂蓝”由耶鲁大学教授卢西亚诺·弗洛里迪(Luciano Floridi)将环境伦理学“伦理漂绿”(Ethics Green Washing)概念类比到数字技术领域提出，指代部分人工智能企业通过市场宣传、广告以及其他公共关系活动，使人们相信他们的行为、产品与服务等合乎伦理原则，但实际情况并非如此的现象。
(1)通常来说，后果论、义务论与美德伦理作为塑造西方道德哲学传统的三大规范理论，共同构成了人工智能伦理原则的哲学基础。但目前研究明显倾向于义务论，并且具有“道德泛法律化”倾向。
(1)很长时间内，教育人工智能中的“公平”倾向于过程公平。1997年，《计算机与教室：美国学校的技术现状》指出，公平是“技术问题”而非“伦理问题”，最好解决策略是为社会经济地位低下的教师和学生提供额外的技术来“创造公平的竞争环境”。(参见Coley R J,Cradler J,et al.Computers and Classrooms:The Status of Technology in U.S. Schools[J].Princeton Nj Educational Testing Service,1997:71.)。
(2)当前教育人工智能伦理原则以集合而非层次结构的形式出现，这“暗示”着各原则要尽可能地实现。但伦理原则想要发挥实际作用则必须相互权衡。如在隐私保护方面，既需要保障学生的隐私和数据安全，又需要充分利用数据进行模型训练和优化，在设计隐私保护策略时，需要考虑不同利益方的需求，做到原则的弹性适应。
(3)例如，正义美德对应公平原则，一个正义的人会“自动地”构建不歧视任何群体的人工智能应用，且不受外部因素的影响；诚实美德对应透明度原则，一个诚实的人会“自动地”公开错误，不隐瞒技术缺陷，使设计的人工智能易于理解和解释；关怀美德对应安全原则，关怀美德将促使人工智能开发者与应用者承认安全与避免伤害的重要性，并采取相应的行动。
(1)根据既定惯例，此类监管机构可以采取多种形式，如数字机构、独立监管机构、检查机构、公共部门监管机构或工作组等。
基本信息:
中图分类号:G434
引用信息:
[1]白钧溢,于伟.教育人工智能伦理原则的有限性及其补正——兼论“原则-美德-法律”框架[J].中国电化教育,2024,No.445(02):23-31.
2024-02-07
2024-02-07