Preserving Trust in Generative AI Scientific Communication

Generative Artificial Intelligence (GAI) introduces new uncertainties to scientific communication without necessarily causing trust breaches. Drawing on semi-structured interviews with 50 multi-stakeholder participants and Gioia qualitative analysis, this study investigates trust preservation—maintaining established trust amid technological disruption. We identify a two-phase process: an assessment phase diagnosing information vulnerabilities, and an active maintenance phase featuring two strategies: creating a shared mental model and balancing autonomy and control. Human-XAI (Explainable Artificial Intelligence) symbiosis, defined through an audit-oriented explainability lens, serves as the enabling mechanism. The study extends AI trust theory and offers practical implications for trustworthy GAI-mediated scholarly communication.