本章小结
美国价值体系中,平等、安全、自由一直占据着中心位置,很容易被误读为同一组价值观跨越时空,贯穿始终。事实上,处在不同经济社会背景中的平等、安全与自由,其内涵也一直在改变。一些价值判断得益于人工智能赋能,正在回归其应有之义,当然它们的负面效应自然也存在被放大的风险。
人工智能在司法领域、职场、教育中的应用,将数据基础中的积弊暴露了出来。机器并没有错,而是“喂养”它的数据里已经包含了不公正的人类实践。因为机器的介入,以算法作为表达的平等价值观日渐客体化,促使业界对相关算法加以讨论和修正。这是人工智能对美国价值观积极作用的一面。但是另一方面,随着人工智能技术的提升,安全的价值内涵也在变迁。20世纪未能实现的军事理想正在成为社会现实,去中心化的私权维护正日益被高度中心化的信息集成与人机结合所取代,人的能动性越来越被力求百分百杜绝人对决策和执行过程的非理性干预的价值取向所取代。
如前所述,美国的价值观由自由主义和控制文化两方面构成。人工智能为控制文化在社会、经济、政治、军事、生活等方面找到了新的应用平台与机会,使控制文化得以二次发育,对效率、精确、标准化和中枢化的要求渗透到日常生活方方面面。控制文化与自由主义之间,力量并非对等,因为前者直接受益于科技的进展,有更多的硬件条件和发生频度支撑;后者则相对软性,倚赖道德自律和慷慨善良。
科技史学者休斯(Thomas Hughes)曾指出,19世纪以来的科技发展是大型科技项目的发展,个体的英雄式的研发固然重要,但是系统的构建则不只包含机械和物件,还涉及社会的组织方式。二战后美国的科技发展超越了早前的大发明家模式,更仰赖权力机构的管理能力对各类科研机构和经济组织的协调与整合。[51]人工智能目前仍不能脱离既有科技应用和社会组织独立运作,因此相比起科技本身,其基础技术的沿革与特性,其在经济社会中实际的应用场景,以及学界和一般社会成员如何对其做出评价,这些方面更值得我们关注。尤其是关于机器的道德观念,究竟是采取边沁式的、更为服务于人工智能使用者的义利观,还是采用康德式的、更顾及普遍性真理和道德的义利观,将成为人工智能开发者们将要一直面对的难题。
至于在一些已经不可逆转的领域,价值观不得不呼应人工智能的发展而发生改变,例如人工智能对就业市场的普遍威胁,以及相应而来的对“勤能致富”这一系列20世纪价值理念的重新解读。人在某些领域的算力全方位败给计算机只是时间问题,这一发展是不可逆转的。虽然美国价值传统中具有相当多抗拒“全民基本收入”这类政策的文化元素,但是年轻一代美国人对于“勤奋”的理解已经悄然改变。勤奋不再是不懒惰,勤奋是越来越讲究战略的、跨学科的、终生式的生存法则,需要在人工智能难以“触类旁通”的领域有技巧地发挥人之为劳动者的价值。至于这种多少带着受迫色彩的勤奋,它的边界在哪里,还没有人知道确切的答案。
[1]Beniger J R The Control Revolution:Technological and Economic Origins of the Information Society[M].Cambridge:Harvard University Press,1989.
[2]Larason J,Mattu S,Krichner L,Angwin J.How We Analyzed the COMPAS Recidivism Algorithm[EB/OL].ProPublica(2016-05-23)[2020-09-07].https://www.propublica.org/article/how-wevanalyzed-the-compas-recidivism-algorithm.
[3]Angwin J,Larson J,Mattu s,Krichner L.Machine Bias[EB/OL].ProPublica(2016-05-23)[2020-09-07].https://www.propublica.org/article/machine-bias-risk-assessments-incriminal-sentencing
[4]Short E.It turns out Amazon's AI hiring tool discriminated against women[EB/OL].Siliconrepublic(2018-10-11)[2020-09-12].https://www.siliconrepublic.com/careers/amazon-ai-hiring-tool-women-discrimination.
[5]Penn State News.Using artificial intelligence to detect discrimination[EB/OL].(2019-07-11)[2020-09-23].https://news.psu.edu/story/580213/2019/07/11/research/usingartificial-intelligence-detect-discrimination.
[6]Short E.3 surprising revelations from IBM's latest HR study[EB/OL].Siliconrepublic(2018-10-02)[2020-09-12].https://www.siliconrepublic.com/careers/ibm-hr-study.
[7]Short E.Diversity deemed top trend shaping the future of recruitment[EB/OL].Siliconrepublic(2018-01-22)[2020-09-15].https://www.siliconrepublic.com/careers/diversityrecruitment-hiring-linkedin.
[8]Stewart D.Control Bias and Eliminate Blind Spots in Machine Learning and Artificial Intelligence[EB/OL].Gartner Research(2018-09-13)[2020-10-01].https://www.gartner.com/en/documents/3889586.
[9]Ruiz R.Facebook's AI suicide prevention tool can save lives,but the company won't say how it works[EB/OL].Mashable(2017-11-29)[2020-04-12].https://mashable.com/2017/11/28/facebook-ai-suicide-prevention-tools/.
[10]Hollister S.Apple is now building the chip it needs to ditch Qualcomm like it ditched Intel[EB/OL].The Verge(2020-12-10)[2020-12-28].https://www.theverge.com/2020/12/10/22168779/apple-leak-cellular-modem-johny-srouji-town-hall.
[11]富兰克林·福尔.没有思想的世界——科技巨头对独立思考的威胁[M].舍其,译.北京:中信出版集团,2019:49.
[12]Spice B.Carnegie Mellon transparency reports make AI decision-making accountable:Figuring Out Why the Computer Rejected Your Loan Application[EB/OL].Carnegie Mellon University,Computer Science Department(2016-05-24)[2020-10-03].https://csd.cmu.edu/news/carnegie-mellon-transparency-reports-make-ai-decision-making-accountable.
[13]国家标准化管理委员会工业二部,中国电子技术标准化研究院.人工智能标准化白皮书(2018版)[EB/OL].中国电子技术标准化研究院(2018-01-24)[2021.02-05].http://www.cesi.cn/201801/3545.html.
[14]Teich P.Artificial Intelligence Can Reinforce Bias,Cloud Giants Announce Tools for AI Fairness[EB/OL].Forbes(2018-09-24)[2020.04-14].https://www.forbes.com/sites/paulteich/2018/09/24/artificial-intelligence-can-reinforce-bias-cloud-giants-announce-toolsfor-ai-fairness/#5b5b3eee9d21.
[15]Taylor F W.The Principles of Scientific Management[M].New York and London:Harper & Brothers Publishers,1911:53-71,12,7-8.
[16]Edwards P.The Closed Word:Computers and the Politics of Discourse in Cold War America[M].Cambridge:The MIT Press,1997:7-15.
[17]RBR Staff.Infographic:The Ethics of Artificial Intelligence[EB/OL].Robotics Business Review(2019-07-12)[2020-07-04].https://www.roboticsbusinessreview.com/ai/infographic-the-ethics-of-artificial-intelligence/.
[18]Schroth L.Drones and Artificial Intelligence[EB/OL].Droneii.com(2018-08-28)[2020-11-15].https://www.droneii.com/drones-and-artificial-intelligence.
[19]Cockburn A.Kill Chain:The Rise of the High-Tech Assassins[M].New York:Picador,2016:225-226,257.
[20]Campaign to Stop Killer Robots.About Us.[2020-08-27].https://www.stopkillerrobots.org/about/.
[21]Lopez C.T.DOD Adopts 5 Principles of Artificial Intelligence Ethics[EB/OL].U.S.Department of Defense(2020-02-25)[2020-08-15].https://www.defense.gov/Explore/News/Article/Article/2094085/dod-adopts-5-principles-of-artificial-intelligenceethics/.
[22]Dillow C.What Happens When You Combine Artificial Intelligence and Satellite Imagery[EB/OL].Fortune(2016-03-30)[2020-08-27].https://fortune.com/2016/03/30/facebook-ai-satellite-imagery/.
[23]Rohn A J.Google's PlaNet:Geolocating Photos Using Artificial Intelligence[EB/OL].GIS Lounge(2016-03-11)[2020-08-27].www.gislounge.com/google-planet-geolocating/.
[24]Etzioni A,Etizioni O.Incorporating Ethics into Artificial Intelligence[J].Journal of Ethics 2017(21):403-418.
[25]Lohr S.Homes try to reach smart switch[N/OL].New York Times(2015-04-23)[2020-11-14].http://www.nytimes.com/2015/04/23/business/energy-environment/homes-tryto-reach-smart-switch.html?_r=0.
[26]Brozek B,Janik B.Can artificial intelligences be moral agents?[J].New Ideas in Psychology,2019(54):101-106.
[27]Wallach W,Allen C.Moral machines:Teaching robots right from wrong[M].New York:Oxford University Press,2010:36.
[28]Allen C,Varner G,ZINSER J.Prolegomena to any future artificial moral agent[J].Journal of Experimental & Theoretical Artificial Intelligence,2000,12(3):251-261.
[29]Arnold T,Scheutz M.Against the moral Turing test:accountable design and the moral reasoning of autonomous systems[J].Ethics of Information Technology,2016(18):103-115.
[30]Gabriel I.Artificial Intelligence,Values,and Alignment[J/OL].Minds and Machines 2020,(2020-01-13)[2020-07-16].https://arxiv.org/abs/2001.09768.
[31]Shank D,Desanti A.Attributions of morality and mind to artificial intelligence after real-world moral violations[J].Computers in Human Behavior,2018(86):401-411.
[32]Shank D,Desanti A,Maninger T.When are artificial intelligence versus human agents faulted for wrongdoing?Moral attributions after individual and joint decisions[J].Information,Communication & Society,2019,22(5):648-663.
[33]杜严勇.人工智能伦理引论[M].上海:上海交通大学出版社,2020:61.
[34]Shaw J.Artificial Intelligence and Ethics:Ethics and the dawn of decision-making machines[J/OL].Harvard Magazine,(2019-01)[2020-08-03].https://harvardmagazine.com/2019/01/artificial-intelligence-limitations.
[35]Patila I,Cogonia C,Zangrandob N,Chittarob L,Silania G.Affective Basis of Judgment-Behavior Discrepancy in Virtual Experiences of Moral Dilemmas[J].Social Neuroscience,2014,9(1):94-107.
[36]Burton E,Coldsmith J,Koenig S,Kuipers B,Mattei N,Walsh T.Ethical Considerations in Artificial Intelligence Courses[J].AI Magazine,2017,38(2):22-34.
[37]Turner,F.From Counterculture to Cyberculture:Stewart Brand,the Whole Earth Network,and the Rise of Digital Utopianism[M].Chicago:University of Chicago Press,2006:31-33.
[38]加里·史密斯.错觉——AI如何通过数据挖掘误导我们[M].钟欣奕,译.北京:中信出版集团,2019:289-291.
[39]布鲁斯·施奈尔.数据与监控——信息安全的隐形之战[M].李先奇,黎秋玲,译.北京:金城出版社,2017:25-28.
[40]Greenwald G.No Place to Hide:Edward Snowden,The NSA,and the U.S.Surveillance State[M].New York:Metropolitan Books,2014:92-94.
[41]Rainie L,Anderson J.The Future of Jobs and Jobs Training[R/OL].Pew Research Center(2014-08-06)[2020-08-16].https://www.pewresearch.org/internet/2014/08/06/future-of-jobs/.
[42]Dillow C,Rainwater B.Why‘free money’could be the future of work(universal basic income as a solution to technological unemployment).Fortune,2017.176(1):69-76.
[43]值得注意的是,即便是60年代全民基本收入呼声高涨时,弗里德曼等经济学家的主要考量仍然在于精简政府机构,削减政府职能,避免形成日益臃肿的福利制度官僚体系,考量的出发点仍然是政府的行政效率,而非对于劳动制度和勤劳这一美德的根本质疑。Friedman M.Capitalism and Freedom[M].Chicago:University of Chicago Press,1962.
[44]Light L.The backlash against Bill Gates'call for a robot tax[EB/OL].CBS News(2017-03-10)[2020-08-15].https://www.cbsnews.com/news/bill-gates-tax-robots-luddite/.
[45]Matthews D.Hillary Clinton almost ran for president on a universal basic income[EB/OL].Vox(2017-09-12)[2020-08-15].https://www.vox.com/policy-and-politics/2017/9/12/16296532/hillary-clinton-universal-basic-income-alaska-for-america-peter-barnes.
[46]Risse M.Human Rights and Artificial Intelligence:An Urgently Needed Agenda[J].Human Rights Quarterly 2019,41(1):1-16.
[47]Goldin I.Five reasons why Universal Basic Income is a bad idea[N].FT.com,2018-02-11.
[48]Touretzky D,Gardner-mccune C,Martin F,Seehorn D.Envisioning AI for K-12:What Should Every Child Know about AI?[M]//Proceedings of the Thirty-Third AAAI Conference on Artificial Intelligence.Palo Alto,CA:AAAI Press,2019:9795-9799.
[49]Touretzky D,Gardner-mccune C,Breazael C,et al.A Year in K-12 AI Education[J].AI Magazine,2017,38(2):5-6.
[50]Cassell J.Artificial Intelligence for a social World:The underlying mission is to better understand social interaction and to build machines that work more collaboratively and effectively with humans[J].Issues in Science and Technology,2019,35(4):29-36.
[51]Hughes T P.Rescuing Prometheus:Four Monumental Projects that Changed the Modern World[M].New York:Vintage Books,1988:6-9.