인공지능 관련 규범 수립의 국내외 현황과 과제

Abstract

Artificial intelligence not only liberated humans from simple repetitive tasks, but also surpassed humans in high-level areas such as Go. The Chat GPT, which was released at the end of 2022, opened a new one of artificial intelligence and triggered the competition for technology development. On the other hand, artificial intelligence has revealed the problem of reflecting discrimination or bias that exists in reality due to the nature of machine learning. It also depicts the future of war weapons with artificial intelligence or surveillance society using facial recognition. International organizations, major countries and companies are striving to prepare norms related to artificial intelligence. The European Union's discussion has been the most comprehensive and consistent, and the 'Artificial Intelligence Act' prepared in April 2021 is expected to have a ripple effect comparable to GDPR. The Korean government is ranked 6th in the world in AI readiness index, but unlike the government sector and data and infrastructure sectors, it is evaluated low in the technology sector. While promoting the development of artificial intelligence technology, it is required to design norms that serve as guidelines and safety devices so that future risks can be prevented and society can enjoy the benefits of artificial intelligence technology development evenly. The establishment of norms for artificial intelligence needs to develop in three directions. First, it is necessary to set and share the values and principles that the nation and society as a whole should pursue. The three principles of AI ethical standards established at the end of 2020 need to be refined into specific practical principles according to the 10 key requirements, and it is desirable to base them on the principles and fundamental rights of the Constitution. Second, preliminary regulations are needed to prevent artificial intelligence from infringing human rights and dignity in public areas. Systems that may cause harm to the body or mind and real-time monitoring systems through facial recognition should be prohibited. Artificial intelligence is useful for benefit administration or crime prevention, but excessive personal information collection, monitoring, and control should be avoided. For automated decision-making systems, the data and algorithms should be checked by humans and notification and explanation to users and the right to opt-out should be legislated. Lastly, the use of copyrighted works for machine learning in the private sector must be legislated as fair use, and the use of disclosed personal information must be allowed without prior consent. Through the introduction of non-identification technology, a legal system that allows data to be utilized without conflicting with personal information protection must be established. Public data, such as court judgments, should be open to the private sector in a usable form. Legislation should be prepared to establish a data ecosystem that allows companies to share, trade, and exchange de-identified data, but it is necessary to prepare measures to prevent illegal activities such as re-identification.