How public discourse on medical AI shapes governance expectations: A Weibo-based mixed-methods study from China

Ting Jiang^1,2*Na Wei¹Qiang Yan^1*Weicheng Ye¹

• ¹Beijing University of Posts and Telecommunications, Beijing, China
• ²Tsinghua University, Beijing, China

Objective: Public perceptions of medical artificial intelligence (AI) directly influence its implementation and governance. While most existing research focuses on Western contexts, there is limited exploration of public responses in collectivist cultures and state-driven healthcare systems like China, particularly regarding the dynamic interplay of cognition, affect, and behavior. This study aims to fill this gap by examining public discourse on medical AI in China, with a specific focus on topic landscape, sentiment distribution, and the Cognition-Affect-Behavior (CAB) mechanisms driving governance. Methods: We collected 12,356 valid Weibo posts on medical AI from January 2022 to December 2024. The Latent Dirichlet Allocation (LDA) topic modeling identified key topics, sentiment analysis assessed emotional tendencies, and grounded theory analysis was applied to 1,000 posts using open, axial, and selective coding to construct a theoretical model. Results: The findings revealed that public discussions covered eight key topics, categorized into three dimensions: foundational drivers of medical AI development, application domains of medical AI, and societal benefits and risks challenges. All topics exhibited a coexistence of positive and negative emotions. The CAB model showed that, cognitively, the public emphasized the human core of healthcare, while acknowledging AI's efficacy, leading to a collaborative augmentation model for the physician-AI integration, where decision-making is physician-led, and AI serves as a supportive tool. Emotionally, the public expressed both amazement at AI's capabilities and expectations for physician-AI integration, alongside resistance to AI and anxiety about the physician-AI integration. Behaviorally, three proactive agency governance strategies were observed, which either reinforced or recalibrated existing cognitive frameworks. Conclusion: This study provides valuable insights into the public's cognitive and emotional responses, as well as proactive behaviors toward medical AI in China. It also highlights the emergence of bottom-up accountability mechanisms, where civic engagement shapes the development of AI governance frameworks in healthcare.