服务机器人领域人工智能伦理风险评估方法的设计与实践  

Design and Practice of Artificial Intelligence Ethical Risk Assessment Methods in the Field of Service Robots

作　　者：李梦薇 徐峰[1] 晏奇 李芳 刘鑫怡 Li Mengwei;Xu Feng;Yan Qi;Li Fang;Liu Xinyi(Institute of Scientific and Technical Information of China,Beijing 100038,China;MEGVII,Beijing 100190,China)

机构地区：[1]中国科学技术信息研究所,北京100038 [2]北京旷视科技有限公司,北京100190

出　　处：《中国科技论坛》2023年第10期74-84,共11页Forum on Science and Technology in China

基　　金：科技创新2030-新一代人工智能重大项目(2020AAA0105301)。

摘　　要：人工智能技术在服务机器人领域的应用将成为缓解人类劳动压力、降低生活成本、弥补服务业人力不足的重要手段,但随之而来的伦理风险不容忽视。本文聚焦服务机器人领域的人工智能伦理风险,首先进行风险识别与风险源分析,进而设计一套通用的人工智能伦理风险评价方法,并在服务机器人这一人工智能技术的重要应用领域进行实践,对服务机器人领域的人工智能决策自主受控风险、隐私侵犯风险、偏见或歧视风险、破坏公平风险、责任归属不清或失当风险这五大伦理风险的大小进行评价,分析各风险的风险源优先级,解决不同伦理风险“轻重缓急”的判断问题,从而验证了该方法的适用性。The application of artificial intelligence technology in the field of service robots will become an important means to relieve human labor pressure,reduce living costs,and make up for the shortage of manpower in the service industry,but the accompanying ethical risks cannot be ignored.This paper focuses on the ethical risk of artificial intelligence in the field of service robots.First,risk identification and risk source analysis are carried out,and then a set of general artificial intelligence ethical risk assessment methods are designed,and practice in service robots,which is an important application field of artificial intelligence technology.Finally,evaluated the five major ethical risks of artificial intelligence decision-making in the field of service robots,privacy violation risk,prejudice or discrimination risk,damage to fairness risk,unclear attribution of responsibility or improper risk,and analyzed the risk sources of each risk.Priority of risk sources for each risk,which solves the problem of“priority”judgment of different ethical risks,and thus verifies the applicability of the method.

关 键 词：服务机器人 人工智能伦理风险 风险识别 风险评价 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]