論人工智慧創作之法律屬性與保護

A Study on the Legal Status and Protection of AIGenerated Work

本文旨在針對人工智慧技術對當下法律之衝擊，以著作權與鄰接權理論為基礎，探討人工智慧創作之法律屬性與保護，以因應法律適用之窘境，尋覓法制與科技之平衡。首先，人工智慧創作之本質係透過演算法對大量作品樣本進行解碼、學習與訓練，形成表徵作品集合之內在邏輯規律的概率模型，並依此進行模仿之結果。人工智慧所習得之作品集內在邏輯與規律可體現著作人對構成元素之原創性選擇與編排，具有可著作性，其著作人是對學習資料、生成模式風格等之選擇與編排具有原創性之人工智慧使用者；其次，對於不具原創性但需要大量資本、技術或勞力投入才能獲得的人工智慧創作，基於保護投資、激勵創造之目的，可參考歐盟資料庫鄰接權之規定，給與15年的鄰接權保護；再次，在法律保護制度上，應當綜合考量多元主體之參與貢獻，並兼顧公共利益，對投資者、研發者、使用者、資料庫者、消費者以及公共大眾進行利益分配，以達到激勵科技創新與進步，增進社會福祉之目的。

"In order to respond to the legal implications caused by artificial intelligence, this article explores the legal status and protection of AI-generated works based on the theories of copyright and neighboring right, tries to solve the dilemma of legal application, and finds a balance between law and science. Firstly, the essence of AI-generated works is to decode, learn and train a large number of samples through algorithms to form a probability model that expresses the inherent logic and rules of the collection. The in-house logic and rules of the collection of works obtained by artificial intelligence can reflect the original selection and arrangement of the elements of the author, which is copyrightable. And the author is the artificial intelligence user with the minimal degree of creativity in the selection and arrangement of learning materials, modes and styles of generation, etc. Secondly, as for AIgenerated works that are not original but require substantial investment to obtain, we can refer to the EU Database Directive 96/9/EC and give a 15-year protection of neighboring right based on the purpose of protecting investment and incenting creation. Thirdly, as far as the legal protection of AI-generated works, considering the contribution of multiple entities and the public interests, this article emphasizes that the rights and responsibilities among investors, researcher, users, digital owners, consumers and the public should be clarified, in order to incent the innovation and development of science, and enhance social welfare."