| 道德人工智能:基础、原则与设计 |
| |
| 引用本文： | 孙福海,陈思宇,黄甫全,伍晓琪.道德人工智能:基础、原则与设计[J].湖南师范大学教育科学学报,2021(1). |
| |
| 作者姓名： | 孙福海 陈思宇 黄甫全 伍晓琪 |
| |
| 作者单位： | 华南师范大学教育科学学院;华南师范大学德育神经科学与人工智能实验中心 |
| |
| 基金项目： | 教育部人文社会科学研究基金项目“卓越教师行动研究的知识创造原理与方法研究”[20YJC880005];2019—2020年度华南师范大学“挑战杯”金种子培育项目“农村小学教师对SQR的接纳调查研究”[19JKKA04]。 |
| |
| 摘 要： | 随着技术的快速进步,人工智能的决策和行为开始从外部设定转向内在发展,智能主体逐渐具有不同程度的自适应能力、决策能力和行为能力,它的自主能力也不断增强。出于道德考量,自主决策和行为的人工智能开始被视为道德主体。因此,传统道德如何在智能技术中自主发挥作用就成为一个必须面对的问题;后果论、道义论和美德论都具有解决此问题的潜力。要构建能够进行道德决策的人工智能系统,需要确保人工智能的自主推理能够受人类社会道德和价值观约束,与人类价值观保持一致,担负起决策“责任”。实现有“道德”的人工智能是一个复杂的系统工程,价值敏感设计、透明度设计及道德学习设计是开发道德人工智能的三种重要方法。
|
| 关 键 词： | 道德人工智能 道德主体 道德代理 机器伦理 机器道德 |
Moral Artificial Intelligence:Foundation,Principles and Design |
| |
| SUN Fuhai,CHEN Siyu,HUANG Fuquan,WU Xiaoqi.Moral Artificial Intelligence:Foundation,Principles and Design[J].Journal of Educational Science of Hunan Normal University,2021(1). |
| |
| Authors: | SUN Fuhai CHEN Siyu HUANG Fuquan WU Xiaoqi |
| |
| Institution: | (School of Education,South China Normal University,Guangzhou,Guangdong 510631,China;Experimental Center of Moral Neuroscience and Artificial Intelligence,South China Normal University,Guangzhou,Guangdong 510631,China) |
| |
| Abstract: | With the rapid progress of technology,the decision-making and behavior of artificial intelligence(AI)begin to change from external setting to internal development.Intelligent subjects gradually have different degrees of self-adaptive ability,decision-making ability and behavior ability,and their autonomous ability is constantly enhanced.For the sake of moral considerations,artificial intelligence,which makes decisions and acts autonomously,has begun to be regarded as a moral subject.Therefore,how traditional morality plays an independent role in intelligent technology has become a problem that must be faced.The theory of consequence and the theory of deontology and virtue all have the potential in this respect.To build an AI system that can make moral decisions,it is necessary to ensure that the independent reasoning of AI can take into account the morality and values of human society,keep consistent with human values,and take on the“responsibility”of decision-making.The realization of“moral”artificial intelligence is a complex system engineering.Value sensitive design,transparency design and moral learning design are three important methods to develop moral artificial intelligence. |
| |
| Keywords: | moral artificial intelligence moral subject moral agent machine ethics machine morality |
| 本文献已被 维普 等数据库收录！ |
|
