Abstract
People increasingly interact with conversational agents based on artificial intelligence (AI) for a variety of tasks. While these communicative AI (comAI) systems often produce responses that are natural and engaging, the information they provide is not always correct and unbiased. This makes it essential to understand how users perceive the credibility of such systems. Drawing on data from four waves of a longitudinal survey (NT1 = 618, NT4 = 485), we investigated how active users’ perceptions of comAI as a tool or social actor and their primary interaction modality (text vs. voice) influence the credibility perceptions of comAI, both cross-sectionally and over time. We found that, within waves, social presence (T1), enjoyment, and perceived intelligence mediated the impact of tool-social actor perception but not modality on perceived credibility. A random-intercept cross-lagged panel model (RI-CLPM) did not support mediation effects at the within-person level over time. Our study contributes to the field of Human-Machine Communication with novel longitudinal insights into the psychological mechanisms that shape users’ credibility judgments of comAI.
DOI
10.30658/hmc.13.5
Recommended Citation
Klein, S. H., & Utz, S. (2026). Tool or social actor? How AI perception and interaction modality influence perceived credibility of communicative AI over time. Human-Machine Communication, 13. 123-154. https://doi.org/10.30658/hmc.13.5
Open Science Badges
Submit Article
This journal supports Open Science Badges
