Label, Debunk, or Nudge? The Effectiveness of AI Decision Makers and AI Decision Assistants in Reducing Perceived Health Misinformation Credibility

As artificial intelligence (AI) systems are increasingly deployed to combat misinformation, understanding how different AI intervention roles influence users’ responses has become an important challenge. This study investigates how AI decision makers (labeling and debunking) and AI decision assistants (nudging) affect perceived misinformation credibility through different psychological mechanisms. An online experiment (N = 123) reveals that AI debunking and AI labeling resulted in similar levels of information elaboration and effectiveness in reducing misinformation credibility for individuals with different levels of need for cognition. However, AI nudging was more effective among individuals with higher need for control, an effect mediated by perceived autonomy. These findings challenge classic persuasion models in AI-mediated contexts, highlighting boundary conditions for scalable AI-mediated persuasion in the context of misinformation correction.