应急伦理治理:叫停研发AGI等暴利创新是当务之急  

Emergency Response Ethics Governance:It is Imperative to Call off R&D of AGI and Other Profiteering Innovations

作　　者：刘益东[1] LIU Yi - dong(Institute for the History of Natural Sciences,Chinese Academy of Sciences,Beijing 100190,China)

机构地区：[1]中国科学院自然科学史研究所,北京100190

出　　处：《自然辩证法研究》2024年第8期132-140,共9页Studies in Dialectics of Nature

摘　　要：以AI大爆发为例,探讨如何阻止带来暴利的高风险创新,分析为何AI发展前景堪忧却蜂拥而上?超过人类的AGI为何成为发展的目标?探讨作为暴利创新的AGI可能引发的伦理与风险问题,阐明应急伦理治理与应急风险治理的必要性与紧迫性,拓展应急伦理治理以克服科技伦理失灵。AGI不是新工具而是新物种,认为AGI引发第四次工业革命是错误的。产生暴利创新的原因有暴利效应、裹挟效应与盲从效应以及经验思维、正效思维、专业思维、侥幸思维、情感思维、从众思维、“亚当·斯密陷阱”等。作为应急伦理治理、应急风险治理的法宝,抢救式治理运用底线思维、贯彻生命至上尊重人权的伦理准则,实行安全问题一票否决等一系列措施。提出AI致善测试,确保人类可持续安全与发展。Taking Al explosion as an example,this paper explores how to prevent high-risk innovation that brings huge profits,and analyzes the reasons why Al is intensively invested despite its unprepossessing outlook,and why ACI,which surpasses human beings,has become a development goal.It also discusses the ethical and risk issues arising from AGI as a profiteering innovation,and elucidates the necessity and urgency of emergency response ethics governance and emergency risk governance,expanding the scope of governance on emergency response ethics to overcome the dysfunction of scientific eth-ics.AGI is not a new tool but a new species,and it is wrong to consider AGI as the trigger of the Fourth Industrial Revolution.The causes for the profitee-ring innovation include WindfallEfect,Coercion Effect,and Herd Efect,as well as experiential thinking,positive thinking,professional thinking,lucky thinking,emotional thinking,conformity thinking,and Adam Smith Trap,etc.As an effective tool for emergency response ethics governance and emer-gency risk governance,rescue governance applies bottom-line thinking,follows out the ethical principle of putting life first and respecting human rights,and implements a series of measures such as vetoes on safety issues.It proposes AI goodness testing to ensure human's sustainable security and develop-ment.

关 键 词：暴利创新 抢救式治理 科技重大风险 科技伦理失灵 AI五要素 亚当·斯密陷阱 AI致善测试 

分 类 号：B82-057[哲学宗教—伦理学]