[2023-04] How generative AI went from innovation to risk? : Discussions in the Korean Public Spheres through the social amplification of risk framework
How generative AI went from innovation to risk? : Discussions in the Korean Public Spheres through the social amplification of risk framework
Sunghwan Kim Moon Soul Graduate School of Future Strategy, KAIST Jaemin Jung Moon Soul Graduate School of Future Strategy, KAIST Abstract Thereare always two sides to technological progress. It creates the seeds of innovation. But it also creates catastrophic risks. An example of this is the controversy surrounding generative AI (artificial intelligence), which has recently gained attention in the marketplace. This study focuses on generative AI and risk. It aims to explore the process of how generative AI, currently the most advanced technology, becomes a risk in the Korean public sphere. To comprehend the social phenomena associated with this, we conducted an analysis of news titles, articles, and comments from the public. Our attention is concentrated on the timeframe from late 2022 to the first half of 2023, during which discussions surrounding generative AI experienced a significant surge in the public sphere. Throughout the analysis period, we extracted news articles that mentioned any of the following keywords: generative AI, generative artificial intelligence, and ChatGPT. To investigate the process of risk amplification or attenuation, we employed topic modeling utilizing the Social Amplification of Risk Framework (SARF). Through this approach, we examined the dynamics of risk perception and communication surrounding generative AI. Notably, by examining the timing of the news stories analyzed in this study, we observed periods characterized by spikes in news coverage. Topic modeling analysis yielded a total of 17 distinct topics across 3,304 news titles and articles. Additionally, we identified 5 topics within 3,748 comments, providing further insights into public discourse related to generative AI. The findings of the study indicate that three specific events acted as risk signals within the Korean public sphere regarding generative AI. There was a discernible difference in the perception of risk between media sources and public comments. The study also revealed that the amplification effect of news media was constrained within the confines of the amplification station. Based on these findings, it is anticipated that future discussions in the Korean public sphere will shift their focus towards the consequences of industry changes, rather than delving into topics such as AI ethics, global regulatory debates, and transformations within the knowledge industry.
Keywords: generative AI, social amplification of risk framework, risk communication