Let’s Chat If You Are Unhappy – The Effect of Emotions on Interaction Experience and Trust Toward Empathetic Chatbots

Abstract Chatbots are now prevalent in obtaining information and executing tasks on behalf of human users. While the design community has paid more attention to streamlining user-chatbot interactions, the existing literature has not thoroughly examined social elements such as emotions and understood how emotions influence user interaction and trust in chatbots. This study proposes to test how participants perceive an empathetic chatbot versus a non-empathetic one under various emotional states (i.e., positive, neutral, negative) when the chatbot facilitates conversations for student advising via an online platform. The study shows the importance of presenting empathetic cues in the design of chatbots or other intelligent agents. The empathetic behavior of the chatbot improves participants’ trust and perception of the chatbot’s performance. The improvement is more salient to people with negative emotions than people who feel neutral or positive. This interaction effect is explained by people’s perceived effort of the chatbot during the interaction with people, even though the chatbot is a technological agent that acts based on algorithms. The results suggest that people attribute human qualities to chatbots in social interaction and highlight the emotional needs of people who experience negative emotions. Therefore, design efforts need to be designated according to people’s dynamic emotional states.