摘要
在生成式人工智能提问模型创作的背景下,传统“工具论”和形式主义“工具论”都不能妥善回应人工智能生成内容(AIGC)的可版权性问题,将“贡献论”作为AIGC可版权性判定的理论基础具有正当性。在理解“贡献论”时,应从静态、结果的“涵摄”视角转换到动态、过程的“法律—技术”视角。“贡献论”中的“贡献”指的是“人类的独创性过程控制”。人工智能提问模型生成AIGC的版权法评价,应围绕版权法保护的贡献源于人类的“提问”,还是源于提问模型产生的“答案”之问展开。如果版权法保护的“贡献”源于答案,会错误释放版权法应当对提问模型激励的讯号;表达应当忠实于思想,提问模型训练来源的不可靠性以及训练结果的不稳定性,导致人类思想不能稳定地预测AIGC,因此,版权法所保护的“贡献”不应来自答案。人类提问对AIGC的生成产生了独创性过程控制,展示从思想到个性化具体表达的过程;保护人类“提问”贡献还有助于缓和版权法上效率与质量的矛盾问题,版权法应依据人类“提问”的贡献来保护AIGC。
In the context of the creation of generative AI questioning models,neither the traditional“instrumental theory”nor the formalist“instrumental theory”can properly respond to the copyrightability of Artificial Intelligence Generated Content(AIGC).It is legitimate to use the“contribution theory”as the theoretical basis for determining the copyrightability of AIGC.When understanding the“contribution theory”,we should transform the perspective from the static,result-oriented“inclusive”to the dynamic,process-oriented“legal-technical”.The“contribution”in the“contribution theory”refers to“human originality process control”.The copyright law evaluation of AIGC generated by AI questioning models should focus on the basic question of whether the contribution protected by copyright law comes from human“questions”or from the“answers”generated by the questioning models.If the“contribution”protected by copyright law comes from the answer,it will mistakenly release the signal that copyright law should incentivize AI questioning models.Expression should be faithful to thought,and the unreliability of the questioning model training source and the instability of the training results lead to the inability of human thought to stably predict AIGC.Therefore,the“contribution”protected by copyright law should not come from the answer.Human questioning exerts originality process control over the generation of AIGC,demonstrating the process from thought to personalized concrete expressions.Protecting the contribution of human“questions”can also help alleviate the contradiction between efficiency and quality in copyright law.Copyright law should protect AIGC based on the contribution of human“questions”.
出处
《南大法学》
2024年第5期143-160,共18页
NanJing University Law Journal
基金
甘肃省2024年高校科研创新平台重大培育项目“数字资产确权系统及其平台研究”(2024CXPT-22)
江苏省2021年研究生科研与实践创新计划项目“新技术革命背景下数据对知识产权法制度的挑战与应对”(KYCX21_1220)资助。
关键词
工具
人工智能
贡献
著作权法
独创性
Tools
Artificial Intelligence
Contribution
Copyright Law
Originality