A Social Identity Extension to CASA: How GenAI Chatbots’ Attitudinal Congruence and Group Identity Impact User Empathy and Issue Support

The growing use of generative artificial intelligence (GenAI) in strategic communication raises questions about how LLM-powered chatbots shape public perceptions and issue support in contentious, identity-relevant contexts. While the computers are social actors (CASA) paradigm offers a compelling account of user responses to AI chatbots based on social cues, it does not explicitly address how different types of identity-based cues shape psychological processing. This study incorporates a social identity lens, proposing that GenAI chatbots’ identity-based cues—personal vs. group-level—trigger distinct processing routes to persuasion. Findings from an online experiment on LGBTQ+ issue advocacy offer initial support for these propositions. At the personal identity level, a proattitudinal (vs. counterattitudinal) GenAI chatbot increased perceived similarity, enhancing empathy through the mechanism of value congruence. At the group identity level, a minority (LGBTQ+) chatbot prompted message elaboration, increasing empathy through the mechanism of minority influence. At both levels, empathy was linked to stronger issue support through attitudes. Implications for the human–computer interaction and persuasion literature are discussed.