인공지능 활용에 따른 법적 책임과 윤리에 관한 동향

Abstract

Due to the nature of 'decoupling' between artificial intelligence and the intention of the designer and 'black box' problem as well, the traditional negligence liability regime is difficult to function properly in imposing civil liability for errors or omissions in opinion of artificial intelligence. Even it is difficult to apply the current Korean Civil Code provisions, such as ‘liability of possessor or owner of buildings or structures’ and ‘employer's vicarious liability’ provisions, ditectly to the civil cases for errors or omissions in the opinion of artificial intelligence. It is desirable to resolve those cases with strict liability or no-fault liability regime, based on objective indicators such as defects rather than negligence responsibility based on subjective indicators such as intention, fault, or negligence. Some further legislations would be required in the near future to resolve civil cases regarding the defects of artificial intelligence, but even prior to those efforts, a method of directly applying or analogizing the product liability law could be suggested provisionally instead. In judging defects in the opinion of artificial intelligence, it should be considered, most of all, whether the best practice recognized in the industry has been faithfully applied in each step in the main procedure of the machine-learning process used in artificial intelligence models. The bias issue of the data and informations used in the machine-learning process is most important consideration as well. In relation to packet based wiretapping in the national defense and security fields, artificial intelligence needs to be operated systematically with a focus on ‘traffic analysis,’ which is likely to minimize the possible infringement of human rights including privacy of the people. Regarding the responsibility and ethics of artificial intelligence, international organizations and non-profit organizations, as well as governments and corporations, have also been developing various ‘ethical principles’ to remove legal blanks and provide legal basis for the use of artificial intelligence. In this paper, the domestic and foreign discussions on the issues above are explored extensively and the applicability of the specific laws of Korea is reviewed in a comparative way.