인공지능 알고리즘 규율을 위한 법제 동향 — 미국과 EU 인공지능법의 비교를 중심으로 —

In this article, I tried to explain the current status of legal regulations to regulate artificial intelligence algorithms in response to their impact on society. Currently, individual regulations are being enforced for some fields, such as self-driving cars and drones. In response to the phenomenon that artificial intelligence is gradually developing and its application is spreading socially, there is an active movement to enact general laws for artificial intelligence discipline. In the meantime, a number of legislative proposals have been proposed in this regard. And now a consolidation of the agreement between the ruling and opposition parties has been prepared in the subcommittee. Comparatively, AI legislation based on risk management is being discussed in the European Union and the United States. The drafted artificial intelligence integration bill is closer to the European artificial intelligence bill than that of United States. The bill's high-risk artificial intelligence classification is focused on the applicable area. However, the nature or risk of an algorithm depends not only on the application area, but also on the purpose of development and controllability of the algorithm. In other words, if it is developed for the purpose of promoting social harm or threatening human life and health from the beginning, it should be prohibited in principle. In addition, if the operating principle of an algorithm is unknown and is beyond the scope of human control, it should be classified as a high-risk algorithm. Since the proposed legislation focuses on the area where algorithms are applied, it cannot be said that it is not able to adequately respond to the risks of algorithms. In addition, there are various methods of responding to high-risk algorithms, including impact assessment, and the method of applying them can be discriminated from strict to flexible responses. In the end, it is necessary to establish a regulatory system that fully considers the risks of AI and embraces its positive aspects. After AlphaGo, the artificial intelligence of Go, shocked our society in 2016, artificial intelligence technology is infiltrating every area of society. In order to accommodate the advantages and minimize the negative impact, individual legal regulations already exist for self-driving cars and drones. Furthermore, the discussion of the AI ​​Act to regulate artificial intelligence in general and comprehensively is gradually taking shape, and an integrated bill was prepared by agreement between the ruling and opposition parties in the National Assembly. It is expected that there will be a lot of discussion in the future before this bill passes the plenary session. In this article, we introduced the AI ​​legal system that has been discussed and legislated in Korea, and pointed out the matters to be noted for successful legislation. I hope that discussions will be more active in the future for the desirable discipline of artificial intelligence, which is even pointed out as the greatest threat to human society.