The impact of authorship and AI attitude on the perception of message credibility and author competence

Generative artificial intelligence (AI) is increasingly changing the way information is conveyed which raises questions about the perception of author competence and message credibility. This study examines the extent to which the type of author (human vs. AI) and the gender of the author (female vs. male) influence the perceived author competence and credibility of a technical text about large language models (LLMs). In a 2×2 between-groups design (N = 219), study participants were randomly assigned to one of four conditions. They were either presented with a female or male human author or with a female or male AI author and rated the competence of the author. Then all participants read the same text about LLMs and assessed its credibility. In addition, the participants’ individual attitudes toward AI were recorded as a potential moderator variable. The results show that perceived author competence did not depend on the author’s gender, but on the author type. Human authors were attributed higher competence than AI authors. This effect was moderated by attitudes toward AI: The more positive the attitude, the smaller the competence advantage of human authors. When attitudes toward AI were taken into account as a moderator, a gender effect was also observed for perceived message credibility: Texts by male authors were rated as more credible by AI-skeptical individuals than texts by female authors. This effect decreased with more positive attitudes. The findings illustrate that the perception of author competence and message credibility is influenced by readers’ individual attitudes toward AI.