When AI fact-checks false short-form health videos: Effects of AI-driven fact-checking on credibility assessment of the videos

Although AI-driven fact-checking has been extensively applied on social media, it is unclear whether presenting AI rather than humans (e.g. social media platform administrators) as a fact-checker will affect people’s credibility perceptions of misinformation. Furthermore, most studies on machine heuristics have mainly focused on the positive machine heuristic and neglected the negative machine heuristic. In this study, a between-subjects online experiment was conducted to address these gaps in knowledge. The results show that presenting AI rather than platform administrators as fact-checkers indirectly led participants to perceive false short-form health videos as more credible via the mediation of the negative machine heuristic rather than the positive machine heuristic. In addition, among participants who initially held a stronger belief in misinformation, the fact-checking provided by AI (vs administrator) generated even more negative machine heuristic, which further led them to perceive the false videos as more credible.