The impact of prior chatbot identity disclosure on customer tolerance of service failures

Purpose Large language models, such as ChatGPT, have made chatbot responses seem more natural, making it more difficult for customers to distinguish that chatbots are non-human. Although chatbots offer advantages in the service industry, their technical limitations can lead to service failures. This study aims to explore how the disclosure of the chatbot's identity during the initial stage of service, before any service failure occurs, affects customer tolerance of service failures. Design/methodology/approach Three psychological experiments were conducted to assess the effects of chatbot identity disclosure on customer tolerance of service failure. Findings This study confirms that the disclosure of the chatbots' identity at the initial stage of service, before any service failure occurs, has a positive impact on consumers' tolerance of service failures. Compared to non-identity disclosure, chatbot identity disclosure increases consumers' perception of chatbots' sincerity and reduces their disgust, thus increasing the consumers' tolerance of chatbot service failures. In addition, the explanation of service failures moderates the relationship between chatbot identity disclosure and customers' feelings of disgust and perceived sincerity. Practical implications This study reveals the importance of chatbot identity disclosure for service companies. Managers should consider a chatbot's ability to handle service failures when deciding whether to disclose its identity. Originality/value This research contributes to the literature on the positive effects of identity disclosure by investigating how the disclosure of a chatbot's identity during the initial stage of service, prior to any service failure, affects consumers' tolerance of service failures.