인공지능의 개인정보 자동화 처리가 야기하는 차별 문제에 관한 연구

Article 10. of the Constitution of the Republic of Korea provides human dignity and the right to pursue happiness, and Article 11. paragraph 1 of the same law states, “All people are equal before the law. No one is discriminated against in all areas of political, economic, social or cultural life by gender, religion or social status.” In each case, freedom and rights are guaranteed around human dignity. Will we be equal before the law and free from all kinds of discrimination during the fourth industrial revolution era with artificial intelligence? The discrimination that data subject receive from automated processing, especially in big data-based artificial intelligence systems, is already deep in our society, but we do not seem to be even aware that we are being discriminated against. Unlike in the past when the mechanisms for processing personal data from personal data collection to deletion were clear, the issue of transparency and fairness caused by so-called ‘black boxing’, where no one knows the process of handling it, has emerged as a new social issue. Although the minimum amount of personal data should be collected and processed transparently according to the principles of relating to processing of personal data, it is becoming virtually impossible for the data subject to exercise its own personal data decision-making authority because it is difficult to know where and how to collect and process its data through automated processing due to advanced technology development. This automated processing can lead to bias and discrimination of data subjects, which can escalate into racial conflict, and further become a barrier to technological development with a vague sense of rejection of artificial intelligence represented by automated processing. However, we should not overlook the fact that artificial intelligence works by human-made algorithms, and research is needed to solve the problem of bias and discrimination in data subjects due to automated processing. In this paper, we will divide the bias of algorithms into the intended and unintended cases of algorithms, and look at domestic and foreign cases of discrimination of artificial intelligence through the definition of discrimination. In order to see how to protect the data subject from discrimination of artificial intelligence, we will focus on the EU’s GDPR from a personal data protection perspective, and look for solutions based on the nature of the algorithm for the opacity derived from the machine learning algorithm, which is currently the mainstream of artificial intelligence.