管理工程学报2026，Vol.40Issue(1)：101-112,12.DOI:10.13587/j.cnki.jieem.2026.01.007
越拟人化越好吗?拟人化AI客服与失败类型对用户抱怨倾向的影响:ERPs证据
Is the more anthropomorphic the better?The effect of anthropomorphic AI customer service and failure type on user complaining tendency:ERPs evidence
摘要
Abstract
With the continuous development of artificial intelligence application technology,intelligent service has emerged as a prominent trend in the development of various industries.However,in the current intelligent customer service deployed in the market,the frequent occurrence of basic errors and poor service quality in AI-driven customer support poses a predicament for enterprises.In an effort to alleviate the negative emotions and reactions of users,the anthropomorphic design of AI customer service has become the foremost preference of enterprises.However,doubts arise regarding the effectiveness of anthropomorphic design in AI customer service when service failures occur.It is therefore an urgent imperative to explore how enterprises can use new technologies while concurrently mitigating the adverse consequences arising from such innovations.Reviewing the existing research,scholars have acknowledged the key role of service failure types and AI anthropomorphic features in service failure.Nonetheless,existing scholarly efforts have predominantly focused on result failures while affording less attention to investigating the combined effect of process failure and result failure,as well as AI customer service anthropomorphism.In addition,in the current research on service failure of AI customer service,scholars have primarily relied on introspective method to measure emotional responses,and capture emotions through retrospective cognitive processing.Such approaches may introduce certain inaccuracies,deviating from users' genuine and immediate emotional experiences during service failures,as such,it may lead to compromised data accuracy.A better understanding of the user emotions during service failures is of profound importance to improving the behavior prediction after such failures. Based on the theory of emotional events,this study delves deeper into the underlying mechanism through which service failure types and the anthropomorphic degree of AI customer service impact user decision-making behavior.This investigation employs a combination of through the combination of event-related potential technology(ERPs)and questionnaire.Using ERPs technology alongside a questionnaire survey method,this study accurately measures users' immediate emotions and behavioral responses in a more refined dimension and high time resolution.The relevant characteristic wave P2 of immediate emotions is employed as a differentiated representation of negative emotions.To meet this end,an EEG experiment of anthropomorphic Al customer service under service failure was designed,and the anthropomorphic AI customer service was combined with the failure type to explore the internal mechanism of users' instant emotion and behavior decision-making.Employing the analysis of variance in SPSS software,this study examines the influence of failure type and personification degree on users' emotions and behaviors,while also exploring the moderating effect of self-construction. The results demonstrate that users manifest varying degrees of negative emotions in response to the service failure provided by the anthropomorphic AI customer service,and then produce different behavioral intentions,which depends on the type of service failure(process vs.result failure)and the level of anthropomorphism(high vs.low),as well as the type of user self-construction.Specifically,anthropomorphic AI customer service and the type of service failure have a significant interaction effect on users' propensity to voice complaints.When users face process failure,compared with low anthropomorphic AI customer service,the service failure of high anthropomorphic AI customer service will trigger a larger P2 amplitude.This,in turn,heightens the willingness of users to express dissatisfaction.For users with failed results,compared with high anthropomorphic AI customer service,service inadequacies attributed to low anthropomorphic AI customer service result in a more pronounced P2 amplitude,resulting in users more inclined to complain.At the same time,users with different personality traits have different perceptions of uncertainty and different attitudes towards errors,and self-construction plays an important moderating role.Dependent individuals,the process failure caused by highly anthropomorphic AI customer service leads to a greater P2 amplitude and complaint tendency of users.When it comes to independent individuals,users have a greater P2 amplitude and complaint tendency for the failure of results stemming from low anthropomorphic AI customer service. In summary,this study elucidates the factors contributing to the inconsistent conclusions of the existing research on the degree of personification's efficacy to alleviate the adverse impacts experienced by users.Theoretically,it extends the current body of knowledge concerning the shortcomings of AI-based customer service interactions.It represents a further application of emotional event theory in the field of service marketing,and helps enterprises better understand the different cognitive and emotional responses exhibited by users to service failure under human-robot interaction.In practice,the research findings are helpful for enterprises seeking to optimize the deployment of anthropomorphic AI customer service to communicate with users with different psychological characteristics and service recovery,and provide important reference value.关键词
拟人化AI客服/服务失败类型/负面情绪/自我建构/事件相关电位Key words
Anthropomorphic AI customer service/Service failure type/Negative Emotions/Self-construction/Event-related potential分类
管理科学引用本文复制引用
朱秋华,孙锐,刘闲月,成汝霞,左嘉嘉..越拟人化越好吗?拟人化AI客服与失败类型对用户抱怨倾向的影响:ERPs证据[J].管理工程学报,2026,40(1):101-112,12.基金项目
教育部人文社科规划项目(20YJA630054) （20YJA630054）
国家社会科学基金项目(22BGL006) Humanities and Social Science Planning Project of The Ministry of Education(20YJA630054) （22BGL006）
The National Social Science Fund of China(22BGL006) （22BGL006）
