摘要
近年来,人工智能技术的突破性发展为人类社会带来了诸多便利,但与此同时也给人类社会带来了不容小觑的风险和挑战,由此涌现了一系列关于人工智能伦理治理的研究成果。目前,主要的治理策略包括建构伦理治理模型和倡导伦理服务理念,本文试图提出第三种策略作为平行方案,即以实验方法论为基础开展人工智能伦理治理。以实验方法开展人工智能伦理治理的重要性就在于能够有效提升识别伦理风险、评估伦理治理框架、化解伦理治理冲突的效能。以实验方法开展人工智能伦理治理包括计算实验和社会实验,前者旨在利用计算机仿真技术对场景进行模拟和测试,而后者则旨在通过社会调查、社会访谈等方法对真实场景中的问题展开研究。这两种实验方法有其各自的实施过程和适用场景,但也存在着在实际开展过程中的困难和问题。总的来说,以实验方法开展人工智能伦理治理,通过预先设计可控实验探索风险影响机制,有助于从根本上提升人工智能伦理治理的效能,为缩小人工智能伦理治理中“是什么”到“如何做”的鸿沟提供解决思路。
In recent years,the breakthrough development of artificial intelligence has brought a lot of convenience to human society,but at the same time,it has also brought risks and challenges that can not be underestimated.Therefore,a series of related research on the ethical governance of artificial intelligence have emerged.For the ethical governance of artificial intelligence,its core task is to narrow the gap between what and how.The current main strategies are presented as two:constructing a model of ethical governance and advocating the idea of ethical services.These two strategies have their respective application domains and have played a positive role in improving governance efficiency.However,there is also a certain ethical vacuum in these two strategies,especially the difficulty in deriving how from what and the difficulty in tracing what from how.This article attempts to propose a strategy for constructing a two-way interaction between what and how,which involves conducting ethical governance in artificial intelligence through experimental methods.As a methodological exploration,the experimental strategy of artificial intelligence ethical governance emphasizes exploring what ethical issues are through experimental means,and then responding to ethical governance through experimental methods.The experimental method proposed in this article aims to build a theoretical framework and provide theoretical support for improving the practical effectiveness of ethical governance in artificial intelligence.The importance of carrying out ethical governance in artificial intelligence by experimental method lies in its ability to effectively improve the efficiency of identifying ethical risks,evaluating ethical governance framework and resolving ethical governance conflicts.The reason why conducting ethical governance in artificial intelligence through experimental methods is feasible is because the experimental methods conform to the characteristics of artificial intelligence ethics,namely indeterminacy,insignificance,and hysteresis,facing the core challenges of ethical governance.Therefore,on the one hand,the goal of conducting ethical governance through experimental methods is to find the causal relationship between ethics and governance.On the other hand,the experimental methods of ethical governance are also more focused on situational governance and flexible governance that analyze specific situations.Experimental approaches to ethical governance in artificial intelligence include computational experiments and social experiments.The former aims to simulate and test scenes with computer simulation technology,while the latter aims to study problems in real scenes through social investigation,social interview and other methods.The design and implementation process of the computational experiment mainly includes:first,proposing experimental hypotheses,that is,conducting experiments on what ethical issues of artificial intelligence;second,conducting abstract designs,setting detailed parameters such as actors,environment,interaction rules,and time scales;third,carrying out experimental operations again,specific simulation programs can be written or some universal simulation platforms(such as NetLogo)can be used to simulate and model the above experimental design.Finally,the experimental results can be analyzed to test whether the experimental hypothesis can be established or whether new experimental conclusions have been obtained.The implementation process of social experiments involves:(1)proposing experimental hypotheses and expected goals,with special attention to measurable goals of ethical impact;(2)selecting experimental platforms,such as online experimental methods;(3)designing experimental methods,that is,by setting up experimental and control groups,designing scenarios and regulating variables to measure indicators and variables related to artificial intelligence,such as scene elements,experimental subjects,experimental indicators,risk prediction and grading,etc;(4)conducting experimental operations,based on experimental hypotheses,by using a combination of online social experiments and situational social experiments,and selecting appropriate subjects through the principle of random allocation to conduct ethical experiments;(5)analyzing experimental results,processing them through measurement and statistical methods,verifying experimental hypotheses,or drawing new conclusions.These two experimental methods have their own implementation process and application scenarios,but there are also difficulties and problems in the actual development process.In general,carrying out ethical governance in artificial intelligence with experimental methods and exploring the risk impact mechanism through pre-designed controllable experiments are conducive to fundamentally improving the effectiveness of ethical governance in artificial intelligence and providing solutions for narrowing the gap from what it is to how to do in ethical governance in artificial intelligence.
作者
于雪
YU Xue(Department of Philosophy,Dalian University of Technology,Dalian 116024,China)
出处
《科学学研究》
CSSCI
CSCD
北大核心
2024年第9期1793-1799,1807,共8页
Studies in Science of Science
基金
国家社会科学基金一般项目(22BZX025)。
关键词
人工智能
伦理治理
社会实验
计算实验
治理效能
artificial intelligence
ethical governance
social experiment
computational experiment
governance effectiveness
作者简介
于雪(1989-),女,副教授,博士,E-mail:yuxue@dlut.edu.cn。
