人工智慧生成作品的著作權法之問

人工智慧時代的機器創作，對著作權法律帶來諸多挑戰。機器創作過程，包括“資料登錄—機器學習—結果輸出”三個階段，分別涉及“機器閱讀”的行為性質問題、“機器創作”的主體資格問題、“機器作品”的可版權性問題。輸入階段的資料採擷，意味著對已有作品自動化、批量化的“閱讀”，是為著作權法中的合理使用；學習階段的“演算法創作”，實際上是“機器作者”與人類作者的共同創作，有別於“人類中心主義”下的創作主體結構；輸出階段的“生成內容”，具有作品的思想表現形式和人格主義要素，應受著作權保護，但其權利由參與創作或投資的自然人或法人行使。面向未來，人類社會應秉持技術理性和制度理性，對人工智慧時代的著作權法之問作出回答。

Machine writing in the era of artificial intelligence(''AI'')has posed a number of challenges to the copyright law. The machine writing process includes three stages, namely data input, machine learning and result output, which respectively involve three copyrightable issues: the nature of ''machine reading'', the subject qualification of ''machine writer'', and the copyright of ''machine-generated works''. Data mining at the input stage means that the automatic and batched ''reading'' of the existing works is fair use. The ''algorithm writing'' at the learning stage is actually a co-creation of ''machine authors'' and human authors, which is different from the structure of creative subject under ''anthropocentrism''. The ''machine-generated content'' at the output stage, with the expressive form of the work and the personality elements, shall be protected by copyright law. However, its rights belong to the natural or legal person who participates in creation or investment of AI systems. In the future, human society should uphold technical rationality and institutional rationality to answer these questions about the copyright law in the era of AI.