Generative AI and Disinformation| Parental Perceptions of Dynamic Exchanges of Human Papillomavirus Vaccine Misinformation and Corrections From AI Checkers on Reddit

Abstract

As vaccination misinformation proliferates online and hesitancy increases in the United States, more research is needed to study how to best correct such myths. This study addresses two issues: exploring corrections from bot versus human actors and testing complicated social media interactions with two misinformation claims about human papillomavirus (HPV) vaccines. In our preregistered survey experiment (N = 1576) of parents, we found that when considering simple corrections to a single misinformation claim, a bot correction increased belief accuracy compared with a control (absent misinformation), whereas a user correction did not. However, as soon as a second, related false claim about the HPV vaccine was raised, audience beliefs about both false claims, as well as attitudes toward the vaccine, were stable, no matter the content and source of the corrections—from a repetitive bot, a responsive bot, or a social media user. We highlight the potential challenges of correction efforts in online environments.