Digital divides in verification: investigating the heterogeneity of verification patterns for AI-generated health information

With the widespread integration of generative artificial intelligence (GenAI) into health information seeking, the prevalence of AI hallucinations necessitates effective verification behaviors. Grounded in the digital divide framework, this study investigates heterogeneous verification patterns for AI-generated health information. We surveyed 597 Chinese adult residents and identified four verifier profiles through the latent profile analysis: verifiers preferring internal verification, verifiers preferring institutional sources, verifiers preferring interpersonal sources and verifiers emphasizing multiple sources. Further analyses suggested that demographic and socioeconomic factors (gender, age, education and income) significantly predicted profile membership. Additionally, the study revealed significant outcome divides associated with individual differences in verification patterns. Specifically, verifiers emphasizing multiple sources reported the highest levels of self-efficacy in identifying AI-generated health misinformation and superior health management outcomes. Notably, verifiers preferring interpersonal sources exhibited weaker self-efficacy and worse health management outcomes compared to verifiers preferring institutional sources. These findings highlight the structural barriers underlying verification divides, offering empirical implications for designing targeted interventions to promote public verification behaviors and narrow associated digital divides.