차이에서 차별:뉴스 데이터 LDA 토픽모델링을 통한 지능정보사회 이슈 분석과 함의

초록

The purpose of this study is to provide implications for discrimination issues in an AI-driven intelligent information society by gaining insights into the meanings of difference and discrimination in our society. To achieve this, the study considered 2010—the year Industry 4.0 (Industrie 4.0) began in Germany—as the starting point of the intelligent information society. It collected news data related to difference and discrimination reported in the media up until June 2024. The analysis was conducted using Latent Dirichlet Allocation (LDA) topic modeling and a literature review on discrimination in the intelligent information society. Through this approach, the study explored and analyzed discrimination issues arising from economic, social, and cultural differences across various aspects of society. Through this study, it was clearly identified how economic disparities lead to discrimination in various areas, including purchasing power, real estate ownership, and access to financial services, ultimately exacerbating structural inequality. Furthermore, differences and discrimination in our society were found to be reproduced through mass culture, sports, and advertising, reinforcing stereotypes and social biases. This has significant implications for discrimination that might be generated by AI-driven tools, which learn from data that reflect traces of human society in the intelligent information age. In a society where differences are perceived not as mere existence but as stereotypes and biases, leading to discrimination, AI is likely to replicate such tendencies or even reinforce discrimination through bias-driven reinforcement learning. To address these issues, the study suggests regulatory measures regarding the documentation of AI training data, based on institutional improvements aimed at fostering an open attitude toward differences and promoting societal awareness changes.

키워드