Using AI chatbots in climate change mitigation: a moderated serial mediation model

Abstract

ABSTRACT This study examined the effect of chatbots’ emotional expression on climate-change-mitigation behaviour intention and the individual and serial mediating roles of social presence and guilt in the main effect. This study further tested the moderating role of custom addressing in direct and indirect relationships between the aforementioned variables. To that end, 577 American adults were recruited through Amazon Mechanical Turk. After eliminating incomplete responses, the remaining 549 participants were assigned to four conditions (factual vs. emotional conditions x control vs. custom addressing conditions), conversed briefly with a chatbot agent, and responded to a questionnaire. The findings showed that chatbots’ emotional expression yielded higher climate-change-mitigation behavioural intention than factual information and that social presence mediated this relationship. The results did not support the mediating role of guilt but supported serial mediation through social presence and guilt. The results further support the moderating role of custom addressing in the indirect relationship through social presence and serially through social presence and guilt. However, the results do not support the moderating role of custom addressing in direct and indirect relationships through guilt.