Conditional effects of source expertise and pre-existing attitudes on objective knowledge in AI-assisted health information verification

Generative AI systems are increasingly integrated into health communication, yet it remains unclear under what conditions source-related cues shape objective knowledge outcomes when individuals verify health information with AI assistance. This study examines the roles of source expertise, pre-existing attitudes, and knowledge states in shaping objective knowledge during AI-assisted health information verification. In an experiment with 103 participants, individuals viewed a mixed-accuracy Facebook post about gluten-free diets attributed to either an expert-labeled or a non-expert-labeled source, then used ChatGPT to verify the information. Source expertise alone did not enhance objective knowledge, but its effect emerged among participants with favorable pre-existing attitudes. The predicted moderating role of knowledge state (uncertain vs. misinformed) was not supported, although exploratory patterns indicated greater responsiveness among uncertain users. The findings suggest that dual-process perspectives are informative for understanding AI-assisted information processing in contexts where verification is supported by generative AI.