인공지능의 신뢰성을 제고하기 위한 저작권법 쟁점 연구

Recently, the use of artificial intelligence, including Chat GPT(Chat Generative Pre-trained Transformer), a generative AI developed by OpenAI, continues to spread in our daily lives. However, the problem is that artificial intelligence technology, which reflects or exacerbates social prejudice and discrimination, from discriminatory natural language processing(NLP) to discriminatory facial recognition related to gender, economic status, education, and race, is applied. As incidents are increasing, the need to correct the errors of discrimination and bias of artificial intelligence and to improve the reliability of artificial intelligence is growing for the sustainable development and develop- ment of artificial intelligence. Because humans are discriminatory and biased, artificial intelligence trained with data that these humans have generated and accumulated for a long time, is also likely to be discriminatory and biased. This is because artificial intelligence learns and studies human thinking, behavioral patterns, and judgment while reading, watching, and listening to human-created works. For this reason, by examining how the copyright law, which protects the creative expression of human thoughts and emotions, can create or promote discriminatory and biased artificial intelligence, how powerful the copyright law is in relation to the development of artificial intelligence in a discriminatory and biased direction. As a way to solve this problem, it is necessary to focus on the issues of discrimination and bias inherent in training data and copyright law issues among the various causes of discrimination and bias errors in artificial intel- ligence. Therefore, it can be seen that there is a possibility that discrimination and bias arising from artificial intelligence can be alleviated and corrected through reasonable application and inter- pretation of copyright law. The reason is that the normative value inherent in fair use, a comprehensive copyright restriction regulation, ultimately coincides with the goal of mitigating errors of discrimination and bias arising from artificial intelligence. Therefore, it is possible to promote the development of fair artificial intelligence through reasonable interpretation and application of fair use regulations under the Copyright Act. Thus, considering the role of copyright law to correct the problem of discrimination and bias errors of artificial intelligence, the most reasonable balance can be established between the protection of copyright holders’ rights and fair use that contribute to the public interest in the era of artificial intelligence.