Transparency, citations, and speed: how design nudges drive adoption in AI health consultations

As explainable AI gains prominence, the question of how its communication design influences user behavior in online health platforms remains underexplored. This study introduces the Human–AI Communication Pathway Optimization Model (HACPOM). This model integrates AI design nudges to explain how transparency of thinking (opaque vs. transparent), source referencing (no source vs. journal), and response latency (fast vs. slow) can influence adoption intention. In a 2 × 2 × 2 between-subjects experiment (N = 723), transparency of thinking and source referencing significantly increased users’ adoption intention. This outcome thus validated the direct route. Response latency showed no direct effect. However, it interacted with transparency, with the strongest advantage occurring under fast response conditions. Structural equation modeling supported HACPOM’s theorized indirect pathway. AI design nudges shaped the perceived risks and benefits, which influenced trust and thereby also influenced adoption intention. A supplementary qualitative interview with Shenzhen residents further corroborated the underlying mechanisms of the HACPOM framework. The implications for designing trustworthy AI health consultations are also discussed herein.