AI for Combatting Mis/Disinformation

Abstract: Artificial Intelligence (AI) poses challenges and opportunities in combating mis/disinformation that arises in today’s digital media landscape. Our research advances understanding of how psychological processing influences perceptions and effects of AI-generated messages on message acceptance and credibility. Specifically, the computers are social actors (CASA) paradigm, anthropomorphism, and solipsistic introjection are employed to examine the effects of the degree of AI involvement in information delivery when seeking clarification on a politicized issue – ranging from none (e.g., news article), to partial (e.g., AI-generated summaries in Google search results), to full reliance on AI (e.g., responses from ChatGPT). Findings from our online experiment with N = 551 US adults indicate that moderate and highly AI-powered sources are deemed more credible than non-AI human-authored sources, however, this effect is weakened by diminished perceptions of anthropomorphizing of AI. Further, solipsistic introjection is negatively related to message acceptance, indicating this intrapsychic processing may decrease the persuasiveness of messages that combat mis/disinformation. Findings are discussed in terms of theoretical and practical implications.