摘要
数据挖掘是生成式人工智能创作的起始环节和必要环节,未经作者许可对作品进行数据挖掘以获取训练数据面临着著作权侵权风险。然而,数据挖掘获取的训练数据是生成式人工智能获得学习能力的重要养料,为化解著作权保护和生成式人工智能技术发展之间的冲突,宜将数据挖掘认定为合理使用。其理论基础在于“过度”的著作权保护已逐渐失去其正当性,而将数据挖掘认定为合理使用,也契合著作权法保护外在表达而非内在思想的原则,而且数据挖掘还有利于实现社会福祉最大化的目标诉求。据此,在弱人工智能时代的情境下,应借鉴域外文本和数据挖掘合理使用制度的先进经验,采用宽松的立法模式创设数据挖掘的合理使用新类型,鼓励生成式人工智能创作,以繁荣相关产业。
Data mining is the starting and necessary step of generative artificial intelligence creation.Without the author’s permission,data mining of works to obtain training data faces the risk of copyright infringement.However,the training data obtained from data mining is an important nutrient for generative artificial intelligence to acquire learning abilities.To resolve the conflict between copyright protection and the development of generative artificial intelligence technology,data mining should be recognized as fair use.The theoretical basis of this is that excessive copyright protection has gradually lost its legitimacy,and recognizing data mining as fair use is also in line with the principle of copyright law protecting external expressions rather than internal ideas.At the same time,data mining is also conducive to achieving the goal of maximizing social welfare.Based on this,in the context of weak artificial intelligence era,it should draw on the advanced experience of foreign text and data mining reasonable use system,adopt a relaxed legislative model to create a new type of reasonable use of data mining,encourage generative artificial intelligence creation,and promote the prosperity of related industries.
作者
周泽中
黎欣
ZHOU Zezhong;LI Xin(School of Law,Hunan Normal University,Changsha 410081,China)
基金
湖南师范大学法学院科研项目(2024HNSDFXY004)
2024年度湖南省法学会法学研究课题青年项目(24HNFX-D-008)。