试论人工智能产品可靠性与企业伦理责任  

On Artificial Intelligence Product Reliability and Corporate Ethical Responsibility

作　　者：褚建勋[1] 刘新生 CHU Jian-xun;LIU Xin-sheng(School of Humanities and Social Sciences,University of Science and Technology of China,Hefei 230026,China)

机构地区：[1]中国科学技术大学人文与社会科学学院,合肥230026

出　　处：《自然辩证法研究》2022年第10期71-77,共7页Studies in Dialectics of Nature

基　　金：中国科学技术大学新文科基金项目“人工智能的基础哲学理论研究”(YD2110002012)。

摘　　要：与用户的关系最为直接和密切,而且作为人工智能产品供应链中的核心企业,在没有满足产品可靠性的情况下就将其投入应用,假如意外发生,按照传统的责任体系,企业应该承担主要甚至全部的道德责任。即使做到了高可靠性,也不意味着人工智能企业就可以被免责,一旦导致危害性的后果,涉及用户的切身利益乃至社会公共安全,便很容易触发人们对企业的反应性态度。那么,根据斯特劳森式的道德责任归因框架,相关企业依然负有一定的责任。总之,相较于传统技术人工物,人工智能产品的特殊之处在于其具有自主性和学习能力,但是对其在可靠性要求上与传统技术物不应存在本质区别。由此表明,人工智能并不必然会导致无法弥合的责任鸿沟。As the core enterprise in the supply chain of AI products,which has the most direct and close relationship with users,it may put the products into application without satisfying the reliability of the products.In case of accidents,according to the traditional responsibility system,the company should bear the main or even all of the moral responsibility.Even if high reliability is achieved,it does not mean that AI companies can be exempted from responsibility,and once AI leads to harmful consequences,involving the vital interests of users and even social public safety,it is easy to trigger people’s reactive attitude towards the companies.According to Strawson’s attribution framework of moral responsibility,related companies should still take certain responsibilities.In conclusion,even compared to traditional technology artifacts,AI products are special in that they have autonomy and learning ability,but there should not be a fundamental difference between these two products in terms of their reliability requirements.This suggests that AI does not necessarily lead to an unbridgeable responsibility gap.

关 键 词：人工智能产品 可靠性 道德责任 供应链核心企业 责任分配 

分 类 号：N031[自然科学总论—科学技术哲学]