How the anthropomorphism of an artificial intelligence agent affects the correction of misinformation: focus on authenticity and behavioural-response system level

AI agents are increasingly being leveraged to counter misinformation. This study investigates their effectiveness in corrections, focusing on the authenticity of user interactions. By analysing anthropomorphism and behavioural responses, it explores how perceived authenticity shapes AI agents' ability to persuade users when correcting misinformation. In a 2 × 2 experimental design involving 320 Korean participants correcting misinformation about the HPV vaccine, highly anthropomorphised AI agents were found to elicit greater perceived authenticity in terms of source, message and interaction compared to low anthropomorphism. Additionally, perceived interaction authenticity mediated the relationship between anthropomorphism and attitudinal change, moderated by the level of behavioural response system. These findings provide vital evidence underlining the importance of anthropomorphism and behavioural responses in fostering authentic interactions, with practical implications.