Trust and Continuance Intention Toward AI-Powered Research Tools: The Moderating Role of Transparency Framing Style

As AI becomes more deeply embedded in scientific workflows, understanding what drives user trust has become central to the sustainable adoption of these tools. This study draws on the Ability-Benevolence-Integrity (ABI) model to examine what shapes trust in AI-powered research tools and how that trust relates to users’ intention to continue using them. A key contribution is the introduction of Transparency Framing Style (TFS) as a novel construct. Whereas transparency refers broadly to how much information an AI system discloses about its data, reasoning and limitations, TFS concerns how that information is structured, emphasised and worded when presented to users. Data were collected from 274 academics and postgraduate researchers across 10 Malaysian universities using purposive sampling and analysed through partial least squares structural equation modelling (PLS-SEM). The structural model explains 68.6% of the variance in trust and 39.8% in continuance intention. Among the ABI dimensions, integrity showed a larger estimated association with trust (β = 0.251, p < 0.001) than benevolence (β = 0.081, p = 0.047). TFS was the strongest direct predictor of trust (β = 0.531, p < 0.001) and positively moderated the relationship between trust and continuance intention (β = 0.180, p < 0.001). These findings extend the human-AI interaction literature by highlighting the importance of interpretive compatibility alongside technical competence in sustaining engagement with AI-powered research tools.