(Im)Balanced customer-oriented behaviors and AI chatbots' Efficiency–Flexibility performance: The moderating role of customers’ rational choices

Artificial intelligence (AI) based chatbots are increasingly deployed in frontline encounters, because they combine frontline service efficiency and flexibility. Using a large-scale data set with more than 130,000 man–machine dialogues from an e-bike sharing platform, Study 1 reveals a complex relationship between chatbots' customer-oriented behaviors and their efficiency–flexibility ambidexterity. Chatbots' level of efficiency–flexibility ambidexterity is higher when their functional and relational customer-oriented behaviors are balanced rather than imbalanced (i.e., a negative imbalance effect) and when they are balanced at a higher rather than a lower level (i.e., a positive balance effect). A follow-up experiment, Study 2, and online survey, Study 3, consistently show that the negative imbalance effect is stronger as customers' perceptions of non-personalization costs decrease and privacy concerns increase, while opportunity cost has no significant influence on the negative imbalance effect. However, consistent with rational choice theory, the positive balance effect is stronger as non-personalization costs increase, privacy concerns decrease, and opportunity cost decreases. In addition, Study 1 and 3 consistently show that in alignment with the stimulus–organism–response framework, efficiency–flexibility ambidexterity partially mediates the relationship between chatbots’ (im)balanced customer-oriented behaviors and customer patronage. This study contributes to the literature on frontline ambidexterity by introducing an AI application context and a more nuanced nonlinear view of the antecedents and consequences of frontline ambidexterity.