From ChatGPT to Sora: Analyzing Public Opinions and Attitudes on Generative Artificial Intelligence in Social Media

Abstract

This study examines public opinions, emotional tendencies, and psychological linguistic characteristics associated with the launch of OpenAI’s ChatGPT and the advanced video generation model, Sora, by analyzing discussions on the Chinese social media platform Weibo. A total of 24,727 valid user-generated texts (1,762,296 words) were collected and analyzed using Python and its associated APIs. Word co-occurrence network analysis, topic modeling based on Latent Dirichlet Allocation (LDA), and emotional characteristics based on the DLUT Emotion Ontology and psycholinguistic analyses based on the Linguistic Inquiry and Word Count (LIWC) dictionary were employed to explore public views on these generative AI technologies. The findings reveal a shift in public focus over time, from initial excitement about technological advancements to growing interest in commercialization, labor, education, ethics, and global competition. The public’s emotional responses to AI were a mix of excitement and apprehension. The study identifies seven distinct emotional types, providing a nuanced understanding of public psychological reactions, which contrasts with previous binary classifications. This research contributes valuable insights for policymakers, businesses, and researchers, highlighting the public’s evolving acceptance of generative AI technologies.

Keywords

• Generative artificial intelligence
• ChatGPT
• sora
• topic modeling
• sentiment analysis
• psycholinguistics