인공지능의 구현을 목적으로 한 저작권법상 텍스트·데이터 마이닝 허용 논의와 관련한 옵트 아웃제 및 조건부 보상시스템 검토

Under the existing copyright law, text and data mining for AI data learning can result in copyright infringement. If greater social benefits can be enjoyed by restricting intellectual property rights in certain cases than without such restrictions, it is reasonable to seek institutional improvement in that direction. Institutional changes that allow text and data mining for developing AI under certain conditions are inevitable. However, AI technology is a disruptive innovation technology that can take away jobs of creators and experts in the future. Authors’ willingness to give up their creations to text and data mining for AI learning purposes can come back as a boomerang that narrows their positions in the future. Even if institutional changes are inevitable in the direction of accepting changed technologies, the current discussion on introducing exemption provisions for text and data mining is problematic in that it presupposes unilateral concessions by authors. Therefore, in this paper, based on the pliability rule that regards the problem of choosing between property rules and liability rules as dynamic and reversible among the previous research results on the property rights system, I argue that it is necessary to give authors a choice between ① declaring an opt-out and thus not being subject to text data mining, or ② allowing text data mining without declaring an opt-out. Regarding the latter choice(②), it is necessary to provide incentives to the author in the direction of remaining without declaring an out-put, through remuneration as a countermeasure for the author’s concession.