Research On The Application Of Fair Use System In Machine Learning

As the core of artificial intelligence technology,machine learning simulates the learning mechanism of the human brain for content creation and output,achieving a leap from human creation to machine creation.It can be said that the process of machine learning is the process of artificial intelligence creation.As an emerging technology,machine learning is now widely used in the research and creation of generative artificial intelligence in the field of literature,but at the same time,this application also faces the risk of copyright infringement.At present,artificial intelligence in China is facing new development opportunities.In order to solve the conflict between copyright protection and social usage needs,and create a good development space for it,incorporating machine learning into the scope of reasonable use is a feasible legal regulatory path.Advocating the application of the fair use system in machine learning and demonstrating its feasibility and legitimacy is an important prerequisite.Given the technical attributes of machine learning and its unique strategic position,it is necessary to standardize its application and development within a reasonable space.Therefore,it is advocated to include it in the scope of reasonable use.On the one hand,the spirit of balancing the interests of fair use and the value orientation of public interest provide feasibility for machine learning to apply the fair use system.On the other hand,from an economic and political perspective,machine learning is justified in applying the fair use system,which can not only effectively regulate market failure in copyright transactions,but also conform to the national policy guidance of promoting the standardized development of artificial intelligence.Based on the current institutional framework,there are still some practical problems in applying the fair use system to machine learning.Firstly,it is difficult to qualitatively characterize the behavior of machine learning in using works at the theoretical level,which leads to the inability to clearly define the scope of the fair use system applicable to machine learning in practice.This can result in a “one size fits all”approach that includes all aspects of fair use or is completely excluded from the scope of fair use.Secondly,the current fair use system cannot provide a legal basis for machine learning.Although the newly revised Copyright Law of the People’s Republic of China(hereinafter referred to as the “Copyright Law”)has added a basic clause on fair use,this clause cannot provide substantive guidance for the application of fair use systems to new technologies such as artificial intelligence.Finally,in judicial practice,there is a lack of unified judgment standards for the fair use system,which leads to too much uncertainty in the application of machine learning to the fair use system.To clear the copyright barriers of machine learning,countries outside the region have sought solutions by improving relevant systems at the legislative and judicial levels.Whether it is the “relatively loose exception model” represented by the United States and Japan,or the “relatively strict exception model” represented by the European Union and the United Kingdom,all have provided useful experience for China to build a fair use system for machine learning.In order to address the challenges posed by new technologies to the copyright system,China urgently needs to reshape the fair use system and provide guarantees for the application of machine learning.Firstly,it is necessary to categorize and characterize the behavior of machine learning using works,and clarify the types of machine learning that can be applied to the fair use system.Secondly,based on the actual situation in our country and drawing on relevant experience from outside the region,we will design a localized system from the perspectives of subject,purpose,and scope of behavior,and add exceptions to the reasonable use of machine learning.At the same time,unify relevant judgment standards in judicial practice,adhere to the “three-step testing method” supplemented by the reasonable use of the “our elements” standard.