Three legal attributes of the hottest AI creations

Three questions on the legal attributes of artificial intelligence creations

up to now, with the catalysis of big data, machine learning, artificial neural network and other technologies, artificial intelligence has more and more human like thinking and creative ability, so it gradually emerges in the field of literary and artistic creation. The challenges brought by the rapid development of technology to the current intellectual property system are also gradually revealed, including whether the AI creations belong to works, how to attribute the rights of AI creations, and what kind of law should be borne by the infringement of AI creations

I. whether artificial intelligence creations belong to works

according to Article 2 of the regulations for the implementation of China's copyright law when the load can no longer rise, works refer to intellectual achievements that are original and can be reproduced in some tangible form in the fields of literature, art and science. No matter from the field or from the perspective of replicability, there is no evidence to deny the status of AI creations. Therefore, originality and intellectual achievements have become the key to judge the attributes of AI creations

first of all, the main reason for the current academic opposition to the originality of AI creations is that AI creations are the result of algorithms, rules and templates. As long as the system itself does not have calculation errors, the generated results will be constant, which cannot reflect the creator's unique experimental force of 50kN/100kN. However, the "unique personality" in the traditional copyright law theory belongs to a very subjective concept. Objectively, as long as the author creates independently and has no substantive similarity with the previous work, it can be judged that the work is original. According to this standard, even if the creation is created by the artificial intelligence system, as long as its creation process is not disturbed by others and the results are different from the previous works, it should be recognized that it meets the requirements of "originality" in the general sense. Secondly, because the current copyright systems in all countries in the world generally focus on human intelligence to build its protection object, the "inhuman" of the creative subject of AI creations has become an important basis for opposing AI creations to become works. But in fact, whether the creative subject is human or not has little to do with whether artificial intelligence creations can be protected as works. On the one hand, from the perspective of the legislative spirit of the copyright law, encouraging innovation is its primary purpose. In this sense, as long as AI creations have the same commercial value as human works, do not constitute plagiarism of previous works, and are innovative achievements produced by using AI technology, they should be protected by copyright law. On the other hand, judging works and non works should be based on objective standards, not on whether the creative subject is human, otherwise it will lead to substantial unfair results. For example, in the case that many AI creations are almost the same as human works and it is difficult to distinguish the true from the false, if they deny their authorship because their creative subject is AI, then in reality, other users can obtain complete copyright, including personal rights and property rights, as long as they hide the creative process and externally show that they are creators. That is, AI creations will be treated differently because of whether to disclose the creative process, which obviously violates substantive fairness. Therefore, the non-human creative subject cannot be the reason to deny the status of AI creations, and the authorship of AI creations should be considered without the influence of the subject

II. How should the rights of AI creations be attributed

the current arrangements for the attribution of the rights of AI creations mainly focus on three types of subjects, namely, AI, AI designers and AI users. In terms of the above three schemes: first, whether through legal fiction or directly giving AI copyright, it will have a subversive impact on the current principles of private law. When artificial intelligence is created by human beings and does not have independent consciousness, its subjective fault and commitment will become a problem, so it cannot become the subject of rights. Moreover, artificial intelligence cannot control wealth, and the incentive function of copyright law has no effect on it, so giving it intellectual property rights is meaningless. Secondly, the reason why artificial intelligence can process and refine the materials it obtains, and finally form achievements comparable to human works, is that developers have contributed to the design of its algorithms and programs. However, in terms of the operation mechanism of AI self generated results, designers cannot control or affect this process from sensing materials, in-depth learning, optimizing models to generating content, and lack substantive contributions to the generated content. Therefore, the creator of artificial intelligence cannot be equal to the right subject of artificial intelligence creations, otherwise it will cause developers to obtain dual rights to an intellectual behavior. Finally, from the perspective of protecting industry and investment principles, giving copyright to AI users will be a more reasonable arrangement. On the one hand, although AI can create independently, its creative intention and the content and form of the final result are still controlled by users. On the other hand, users have made corresponding investments to obtain the right to use the intelligent system. If the benefits brought by its products cannot be occupied by users alone, consumers' interest in AI products will inevitably be reduced, and the development of AI industry will be hit. The possibility of users operating programs in unpredictable ways by programmers to obtain innovative results will also be reduced, which is not conducive to the progress of overall social innovation. Therefore, under the principle of private law autonomy, it is more conducive to enhance the rationality of the benefit distribution of AI creations that there is an agreement to follow its agreement, and there is no agreement to empower users

III. what kind of law should be borne by infringement of artificial intelligence creations

according to the provisions of the current copyright law, the infringer shall bear the corresponding civil, administrative or criminal liabilities according to the situation. Among them, civil, as the most commonly used form that can make up for the loss of the obligee, mainly includes "stopping the infringement, eliminating the influence, making an apology and compensating for the loss". However, the uncertainty of the content of AI creation rights leads to obstacles in the civil application in this field. In this regard, we should first make it clear that AI creations are at least property information. Taking Microsoft Xiaobing as an example, the publication of his poetry anthology "the sun lost the window" can not only attract readers' attention by measuring the width of the sample with a vernier caliper, but also help to enhance the popularity of Microsoft enterprises, establish a professional image, and expand potential trading opportunities. It can be seen that AI creations are no longer the objective result of independent operation and calculation of intelligent systems compared with carbon fiber composites, but have become information products with economic value. In addition, AI creations usually conform to the appearance and form of works, and their contents are "distinguishable differences" from previous works, which can be controlled exclusively for human beings, and have the conditions to become the object of property rights. Therefore, the subject who infringes on AI creations must bear the legal consequences of stopping the infringement, eliminating the impact and compensating for the losses, so as to make up for the property interests damaged by the obligee due to the infringement. Secondly, there is no consensus on whether artificial intelligence works set personal rights. For example, the United Kingdom claims that there is no personal right in computer-generated works in Article 79 (2) (c) of its copyright, design and Patent Act. Therefore, it is still controversial whether the law of "apology" is required for infringement of AI creations. However, it can be made clear that the purpose of this form is to maintain the personal dignity of the infringed. If the infringed subject will not suffer mental pain due to the infringement, for example, it is a legal person organization, the infringer will not need to bear an apology

with the advent of the era of artificial intelligence, changes triggered by new technologies are constantly changing people's production and lifestyle. On the objective level, AI creations have appeared in large numbers and have been used for commercial purposes, becoming a transaction object with economic value in fact. Facing this reality, the current legislation must clarify the legal attributes of AI creations as soon as possible, properly handle the ownership of their rights, and improve their right protection system. Only in this way can we reduce the impact of AI creations on the existing copyright order, avoid the damage to the rights and interests of corresponding creations due to their unclear legal positioning, give full play to the incentive function of the intellectual property system, and promote the development of AI industry and the innovation of cultural, artistic and scientific works

(author unit: School of economic law, Southwest University of political science and law)