A scoping review of trust in firsthand and secondhand communication involving humans and artificial intelligence

Abstract

Conversational and generative artificial intelligence (AI) systems increasingly create and pass information between people. This can make it hard to see who created a claim, whether it was changed, and whom to trust. Many studies examine firsthand human-human and human-AI communication, but less is known about secondhand communication, in which AI or another actor passes on a claim made elsewhere. In these cases, a human decision maker must evaluate the claim and the people or systems that created or passed it on. We present a scoping review built around six minimal motifs that represent firsthand or secondhand communication routes. Humans or AI systems may serve as the source or intermediary, while the decision maker remains human. We screened 3810 records and coded 124 studies to identify 21 trust factors in four dimensions: Decision Maker, Messenger, Message Property, and Environment. Qualitative synthesis also identified four recurring route-level patterns: distorted trust calibration, trust transfer, extremity amplification, and accountability diffusion. These patterns guide future research on AI-mediated communication. Overall, this review provides shared terms and a route-based framework for studying trust in multi-party human-AI communication and designing AI-mediated communication systems.