Exploring Public Sentiments on AI in Mainland China: A Study of User-Generated Data from Weibo

The motivation of this study stems from the increased attention to AI ethics and public opinion since the launch of the large language model product ChatGPT. While international studies often focus on public opinion research on English social platforms, discussions and analyses of AI among mainland Chinese users are infrequently conducted, despite China’s leadership in global AI development. This study aims to bridge this gap by analyzing user-generated data on Weibo, providing insights into the emotional polarity and key discussion topics surrounding AI among Chinese netizens. An ERNIE-BiLSTM was implemented for sentiment classfication, and a LDA-CorEx method was deployed for topic modeling. The sentiment trajectory analysis results shows a post-COVID-19 optimistic rebound. through Latent Dirichlet Allocation (LDA) and review with Correlation Explanation (CorEx), we identify seven key discussion topics, including National Strategy, Entertainment, Robotics, and others, and analyze their sentiment distribution. The findings have implications for policymakers, practitioners, and international stakeholders interested in promoting AI technology and products.