“I Know” or “I Feel”? The Impact of Generative AI Applications’ Empathic Response Types on Customers’ Willingness to Forgive

As Generative AI (GenAI) increasingly functions as a conversational service agent, human–GenAI interactions involve socioemotional cues such as empathy. Whether customers distinguish qualitatively different forms of GenAI-generated empathy and how these cues shape responses to GenAI service failures remains unclear. Four studies examine how GenAI empathic response type (emotional empathy response vs. cognitive empathy response) shapes customers’ willingness to forgive. Emotional empathy responses (vs. cognitive empathy responses) increase willingness to forgive (Studies 1a and 1b). The effect is stronger for high-severity failures but attenuated for low-severity failures (Study 2a). Study 2b uses event-related potential (ERP) analysis to provide neurophysiological evidence of differential customer responses to GenAI empathic response types. Customers’ emotional experience underlies these differential responses; service failure severity is an important boundary condition. This research deepens understanding of socioemotional interactions in human–computer interaction, showing that customers respond differently to GenAI-generated empathic cues despite GenAI’s lack of genuine emotional experience.