논문 제목(국문) : 유럽연합 집행위원회의 인공지능 규정(안)과 윤리
논문 제목(영문) : Proposal for a Regulation of EU Commission on AI and Ethics
저자 : 조동관
소속 : 국회도서관 법률정보실 법률자료조사관
요약 : On the day of 21 April 2021, a proposal for a regulation of EU Commission was published. The proposal suggested that regulation on AI should be regulated by 4 levels based on risk degrees. The actual contents of the regulation are, however, focused on the prohibited AI and the high-risk AI. This paper explains the contents of the prohibited AI and the high-risk AI regulations shortly and examines what relations between the Regulation and ethics are in the application of the Regulation. For this review, 3 cases including AI are supposed in this paper and are used for finding what substantial standards are in the Regulation. The Regulation suggests 4 types concerning the prohibited AI but there are no standards for separations between the prohibited AI and the high-risk AI. The Regulation requests an operator of the high-risk AI get a license on its AI from an authorized institution in his/her country and take a ‘conformity assessment’ of his/her AI before AI services go on. The article 10 of the Regulation describes several standards for taking exact and non-bias data-sets but the contents of the standard are about procedures for getting a license overall. To find what substantial standards are in the application of the Regulation, this paper applied the Regulation to 3 cases and found out the next. First, the substantial standard in the application of the Regulation would be ethics. Second, the substantial standard of the prohibited AI requires risks as very high as criminal level. Third, the substantial standard of the high-risk AI requires risks as high as not over general ethics.
목차 : 1. 서 론
2. 안전한 인공지능과 윤리
3. 결 론