Verifying through involvement: unveiling the potential of anthropomorphism in enhancing users' intentions to verify AI-generated information

The prevalent use of generative artificial intelligence (AI) in everyday life has caused widespread misinformation. Despite the pressing need for users to verify AI-generated content, factors influencing their intentions to do so have remained underexplored. This study, grounded in the involvement theory, examines how AI anthropomorphism affects users' intentions to verify AI-generated content. Through a survey (Study 1, N = 200) and an experiment (Study 2, N = 210), we found that AI anthropomorphism enhanced users' verification intentions both in everyday use and in the experimental setting. This effect was serially mediated by reduced psychological distance (i.e. relational involvement) and increased perceived self-relevance (i.e. cognitive involvement). Moreover, these underlying mechanisms were amplified by users' cooperative mindset towards AI, while attenuated by the competitive mindset. The findings offer new insights into understanding the mechanisms of verifying AI-generated content and have practical implications for optimising AI design.