Abstract
As users increasingly turn to Generative AI (GenAI) for healthy lifestyle advice, they face critical risks of hallucinations and automation complacency. To address this, this research integrates the Stimulus-Organism-Response (S-O-R) paradigm with the Theory of Planned Behavior (TPB) to investigate users’ defensive verification behaviors. We conceptualize GenAI Information Quality and GenAI Explainability as parallel external stimuli shaping internal cognitive evaluations (Trust and Perceived Risk) and motivational states (TPB constructs), which subsequently drive Information Verification Intention and societal-level Recommendation. Using data from 299 U.S. GenAI users analyzed via PLS-SEM, the findings reveal a double-edged reality. High-quality information successfully builds trust and mitigates risk. However, explainability acts as a “double-edged sword”: while it builds trust, highly detailed explanations simultaneously induce “cognitive offloading,” significantly reducing users’ attitude, subjective norm, and perceived behavioral control toward verification. Furthermore, excessive trust inadvertently breeds “automation complacency,” acting as a strong inhibitor to verification behaviors. Ultimately, personal verification intention is primarily driven by attitude and perceived behavioral control, whereas recommendation behavior is strongly driven by subjective norm. Furthermore, perceived behavioral control positively moderates the translation of individual intention into interpersonal recommendation. This study provides evidence-based guidance for responsible GenAI interface design, emphasizing dynamic trust calibration and cognitive forcing functions.
Recommended Citation
Li, Eldon Y.; Huang, Guan-Wei; and Hong, Rui-Yan, "The Effects of GenAI Information Quality and Explainability on User Trust, Perceived Risk, Information Verification Intention, and Recommendations" (2026). ICEB 2026 Proceedings (Beijing, China). 96.
https://aisel.aisnet.org/iceb2026/96