Understanding Public Perceptions of Generative AI: A YouTube Comment Analysis

Generative AI refers to artificial intelligence models that can generate new data by utilizing patterns and structures learned from existing data (Ray, 2023). The emergence of these technologies has raised important questions about their impact on society and has sparked varied discussions, debates, and opinions among the public. YouTube, one of the largest video-sharing platforms on the internet, serves as a rich and dynamic space for people to share their views, engage with content, and interact with others. Consequently, YouTube comments offer a valuable window into the public's collective consciousness, revealing insights into their perceptions, concerns, and attitudes towards generative AI. This research aims to investigate the behavioral aspects of public perceptions of generative AI (e.g., ChatGPT), with a particular emphasis on the discourse found in YouTube comments. The primary objectives of this research are: 1) To identify and analyze the predominant themes and concerns expressed by individuals in YouTube comments related to ChatGPT and generative AI. 2) To examine how the attitudes and perceptions of individuals regarding ChatGPT and generative AI affect individuals’ adoption. 3) To explore the role of communication channels and social systems in shaping the online discourse surrounding AI, and how they influence the diffusion of AI technologies in public perceptions, based on the Diffusion of Innovations Theory (Rogers et al., 2014). We selected a set of YouTube videos that discuss ChatGPT or generative AI, and extracted around 67,000 valid comments from these videos, ensuring a diverse range of sources and perspectives. Data was cleaned by removing spam, irrelevant comments, and any data inconsistencies. Both qualitative and quantitative analysis were conducted. We utilized natural language processing (NLP) tools to perform sentiment analysis on the comments and conducted a thematic analysis to identify recurring themes and concerns among commenters. Sentiment analysis results show 46.9% positive, 27.5% neutral, and 25.6% negative among the cleaned comments. From thematic analysis, we extracted the top 30 topics among the comments. We will also examine the influence of specific YouTube channels or influential figures on shaping perceptions. Finally, we will construct a theoretical model based on the Diffusion of Innovations Theory, to investigate the factors influencing public perceptions of AI in online comments and assess the potential impact of these perceptions on the adoption and regulation of AI technologies. By analyzing the comments posted on YouTube videos that discuss Generative AI, this study attempts to uncover patterns, themes, and sentiments within the online discourse, providing valuable insights into the dynamics of AI technology adoption and acceptance. The significance of this study lies in its potential to shape policies and guide future advancements in AI, enhance public awareness, promote ethical AI adoption, and support decision-making.