The Influence of Framing, Domain and Task Type on Trust in AI
Stefano Guidi, Lorenzo Reina, Francesco Currò, Enrico Cipriani, Simone Grassini, Oronzo Parlangeli · 2025
As artificial intelligence (AI) technologies become increasingly integrated into decision-making processes, understanding how users develop trust in these systems is essential.While prior research has examined isolated factors such as transparency or system performance, less is known about how trust is shaped by the combined influence of narrative framing, task type, application domain, and individual user characteristics.This study investigates these dimensions through a preregistered online experiment (N = 280) employing a 3 (narrative: human-centred, technical, control) x 2 (task type: generative, recommendation) x 3 (domain: healthcare, insurance, relationships) mixed factorial design.Participants were presented with a narrative introduction to a hypothetical AI system and asked to evaluate its trustworthiness, perceived utility, and likelihood of adoption across multiple scenarios.Results showed that narrative framing alone had minimal impact on trust, suggesting that brief descriptions may be insufficient to shape user attitudes.In contrast, both domain and task type significantly influenced trust: AI systems used in insurance and healthcare were trusted more than those in personal relationships, and recommendation systems consistently outperformed generative ones in user evaluations.These findings highlight the importance of contextual and individual factors in fostering appropriate trust in AI.