Health misinformation continues to spread widely on social media. Traditional correction methods seem to struggle to keep up with the speed at which misinformation is disseminated. Chatbots have shown potential in addressing this issue, yet few studies have examined their effectiveness in correcting health misinformation, let alone the persuasion strategies them. This study aims to fill this gap by conducting three experiments to examine the effectiveness of chatbot debunking and identify the most persuasive strategies in combating health misinformation.
The first experiment evaluates whether chatbot interventions significantly improve misinformation identification compared to a control group. The second experiment explores four key persuasive strategies - message sidedness, persuasion appeal, conclusion explicitness, and fear appeal intensity - to determine which approach yields the best corrective outcomes. The third experiment compares chatbot-led debunking to human expert interventions, assessing their relative effectiveness.
Findings across the studies reveal that chatbots significantly enhance misinformation correction, with two-sided messages, rational appeals, explicit conclusions, and high fear appeals proving to be the most effective strategies. However, human experts still outperform chatbots in overall corrective effectiveness.
These findings offer new insights into the media evocation paradigm by highlighting the conscious negotiation process in human–chatbot interactions during misinformation correction. They challenge the traditional computers-as–social-actors paradigm and provide a theoretical basis for understanding chatbots as hybrid agents between humans and machines.
