聊天机器人服务失败中拟人化对顾客反应的影响  

Effect of the Anthropomorphism on Customer Response in Chatbots Service Failure

作　　者：曹忠鹏[1] 马慧楠 严兴全 CAO Zhongpeng;MA Huinan;YAN Xingquan(School of Business Administration,Northeastern University,Shenyang 110169,China;Beihai Vocational College of Wellness,Beihai 536007,China)

机构地区：[1]东北大学工商管理学院,沈阳110169 [2]北海康养职业学院,广西北海536007

出　　处：《管理科学》2023年第1期106-118,共13页Journal of Management Science

基　　金：国家自然科学基金(71672028)。

摘　　要：随着人工智能等技术水平的提高,以及服务行业人工成本的上升,越来越多的企业开始使用聊天机器人作为新型服务方式补充甚至替代一线服务员工。与员工服务一样,聊天机器人服务失败同样不可避免,但鲜有研究探究聊天机器人服务失败中拟人化对顾客反应的影响及其内在机理;此外,已有研究发现人机信息沟通中机器人身份披露和澄清都影响服务失败后顾客的反应,但是预警能否起到一定作用还不得而知。基于社会反应理论和期望不一致理论,考察聊天机器人服务失败中拟人化程度对顾客反应的影响,以及负向不一致的中介作用,根据预防接种理论,引入预警作为调节变量探讨其调节作用。通过预实验成功操控聊天机器人的拟人化程度,然后进行3个实验收集数据,并进行实证分析。利用Spss 22.0进行单因素方差分析,检验拟人化程度对顾客满意度的影响,采用线性回归分析检验满意度对继续使用意愿的影响;采用Process插件和bootstrap分析检验负向不一致的中介作用和预警的调节作用,并尝试排除其他可能的中介解释。实验1的结果表明,高拟人化带来较低的顾客满意度,较低的顾客满意度带来低的继续使用意愿;实验2的结果表明,负向不一致在拟人化程度与顾客满意度之间起中介作用,并排除了服务失败归因和感知有用性的替代中介;实验3的结果表明,预警在拟人化程度与负向不一致之间起调节作用,相对于没有预警,有预警能弱化拟人化程度对负向不一致的影响。研究结果在理论上证明了拟人化聊天机器人对顾客反应的负面影响并揭示了潜在的机制和边界条件,丰富了拟人化的相关研究,尤其是拟人化的消极影响;为探究聊天机器人服务失败下人机信息沟通提供了新的研究视角,拓展了预防接种理论在服务机器人领域的应用范围;为商家设计聊天机器人外观以及�With the improvement of artificial intelligence and other technologies and the increase of labor costs in the service industry,more and more companies have begun to use chatbots as a new service mode to supplement or even replace frontline service employees.Similar to human services,service failures of chatbots are also inevitable.Yet few studies have explored customer responses to failures in chatbot service.In addition,previous studies have found that both identity disclosure and clarification in human-machine information communication can influence customer response,yet whether forewarning can play a role is stillunknown.Based on the social response theory and the expectation-disconfirmation theory,this research investigates the effect of anthropomorphism on customer satisfaction and intention to continued use in the chatbots service failure and the mediating effect of disconfirmation.From the perspective of inoculation theory,we introduced forewarning as a moderator and examined its'moderating effect.The different levels of anthropomorphism of chatbots were manipulated successfully through a pretest,and then we executed three experiments to collect data and analyzed these data to verify these hypotheses.Spss22.0 software and one-way analysis of variance are conducted to test the influence of the level of anthropomorphism on customer satisfaction,linear regression analysis was used to examine the impact of satisfaction on the intention to continued use.Process plug-in and bootstrap analysis were used to investigate the mediating effect of negative disconfirmation and the moderating effect of forewarning and try to rule out other alternative mediating explanations.The study results show that after the chatbots service fails,high(vs.low)anthropomorphism causes lower customer satisfaction,which leads lower intention to continued use(study 1);negative disconfirmation plays a mediating role between the anthropomorphism and customer satisfaction the alternative mediations of locus of attribution and perceived useful

关 键 词：聊天机器人 拟人化 负向不一致 预警 服务失败 

分 类 号：F713.5[经济管理—市场营销]