Is cute AI more forgivable? The impact of informal language styles and relationship norms of conversational agents on service recovery

Conversational agents are increasingly handling a significant volume of service interactions and are projected to handle up to 95% of customer service interactions. However, like any service provider, conversational agents are not immune to service failures, which can result in negative customer experiences. Within the context of interactions between conversational agents and consumers, the informal linguistic strategies employed by conversational agents have been recognized as vital in recovering customer satisfaction after service failures. Nevertheless, research on how these linguistic strategies impact customer responses following service failures is still in its early stages. This study delves into the informal language styles used by conversational agents after service failures, examining the comparative effects of cute language styles (as informal language styles) and literal language styles (as formal language styles) on service recovery. Drawing upon the mind perception theory and the stimulus–organism–response (SOR) model, this research also unveils the underlying dynamics of whimsical and kindchenschema cuteness on customers' forgiveness, along with literal language styles, in the context of service recovery. Furthermore, this study investigates the moderating influence of exchange and communal relationships, taking into account the boundary effects of relationship norms. To explore these hypotheses, three experiments were conducted. The findings of this research offer valuable theoretical insights and practical implications for designing linguistic strategies employed by conversational agents.