Algorithm characteristics, perceived credibility and verification of ChatGPT-generated content: a moderated nonlinear model

Purpose The use of ChatGPT has raised concerns about information credibility, prompting users to engage in verification behavior. The study distinguishes algorithm accountability from algorithm fairness and transparency, examining their effects on perceived credibility and their interrelationships. It further explores the nonlinear relationship between perceived credibility and verification behavior, considering social cues as a moderating factor. Design/methodology/approach Data were collected via an online survey of 208 ChatGPT users. The hypotheses were examined through moderated multiple regression analysis and structural equation modeling. Findings The results show that social cues have a moderated effect on the nonlinear relationship between perceived credibility and verification, showing an inverted U-shaped pattern when social cues are lacking, and a less significant relationship when social cues are abundant. Algorithm fairness and transparency positively affect perceived credibility while algorithm accountability does not. Algorithm accountability has positive effect on algorithm fairness and transparency. Originality/value This research advances the theoretical understanding of human–ChatGPT interaction by exploring the nonlinear relationship between perceived credibility and verification behavior in the context of ChatGPT, revealing the moderating role of social cues. It also extends the fairness–accountability–transparency framework by distinguishing between developer- and technology-related algorithm characteristics and examining their interrelationships.