Trust in AI-Generated Scientific Evidence: Effects of AI Disclosure Statements

Recent research has shown that message credibility is reduced when the respective message was generated by artificial intelligence (AI). However, research on whether this is also true for scientific research is scarce. Considering that researchers increasingly use AI in their work, I examine whether scientific evidence generated with the help of AI is trusted less compared to scientific evidence generated without AI use. The present study will use combinations of author and study descriptions to experimentally manipulate AI use. The studies will thereby focus on the topic of acupuncture research. Study and author descriptions will be experimentally varied regarding the degree of AI use by the respective authors and in the respective studies. Trust in the study will be the main dependent variable, complemented by a number of additional dependent variables such as belief updating or perceived study objectivity. Several potential moderating and mediating factors will be tested in an exploratory fashion.