The problem with AI generated art
While the development and use of tools allowing the generation of images by artificial intelligence have multiplied in recent years, including Midjourney, Stable Diffusion, Dall-E, DeepAI and many others, ethical, legal and philosophical issues arise.
To generate these images, these tools use powerful artificial intelligence technologies, in particular machine learning and deep learning. These programs collect billions of images found on the Internet, analyze them, classify them and give them a description. The more information the system compiles, the more efficient it becomes, it is a real learning mechanism. Quickly, the algorithm is able to recognize the recurrences between the different images analyzed, to reproduce them, and then to combine them to form a new image according to the user’s requests. For example, an algorithm trained on images of paintings will be able to produce entirely new works based on its training data.
The emergence of this new form of AI-driven creativity is already having a profound impact on artificial intelligence technologies, public awareness as well as the creative arts industry. Indeed, many art galleries and museums around the world are already investing in exhibitions specifically dedicated to computer-generated artworks. Professionals of the sector such as graphic designers, draftsmen or artists in the broad sense have started more and more to use these kinds of tools during their creative process. Businesses are not left out, as many of them use AI for design purposes, whether it is to generate logos or to create entire fashion collections from (almost) nothing.
Nevertheless, this growing popularity is accompanied by many controversies. First, the fact that for the first time, an AI won an art competition, in this case, Midjourney. It is indeed Jason M. Allen, a video game creator from Colorado, who won on August 22, 2022 the Colorado State Fair Fine Arts Competition in the category “digital art / digitally manipulated photography” using the Midjourney tool. This was his first time ever entering a competition of this kind. The work entitled “Théâtre d’Opéra Spatial” is controversial since many critics question its validity because Jason M. Allen did not create the image himself since he used Midjourney.
The artist defends his victory and mzintains that he respected the rules of the contest: “I won, and I didn’t break any rules”. The rules of this category allow any “artistic practice that uses digital technology as part of the creative or presentation process”. The two judges have since announced that they did not know what Midjourney was at the time of their deliberations, but they said that they would have awarded the first place to Jason M. Allen nonetheless.
Some people have also denounced the little amount of work and know-how that the use of such tools requires in contrast to the so-called “traditional” methods, believing that pressing a few buttons is not enough to call this work ‘art’. On this subject, the artist defends himself by stating that he worked more than 80 hours in total on Midjourney to reach the final result. Much more work, then, than just pressing a few buttons.
IA-generated art also raises important legal issues related to intellectual property rights. Indeed, as previously explained, in order to function, these tools analyze and rework a huge quantity of images and works that are free of rights, but also of authors and artists who have generally not given their authorization for such a practice. Moreover, it would seem that some of the images generated by these AIs may be substantially similar, or even identical, to works created by these traditional artists, so the question of plagiarism and copyright arises.
In recent months, there have been several high-profile disputes between creators and AI companies claiming that some of their works have been copied by AI systems without permission and without crediting the original author of the work. These disputes raise many questions about who owns the copyright to AI-generated works — is it the artist whose work was analyzed/copied? Is it the user of these platforms who entered the instructions? Is it the software publisher? Or is it the algorithms themselves who might hold the copyright on their own creations?
To face this problem, more and more artists are approaching associations or professional unions to assert their rights or to create support communities on social networks. The movement “SAY NO TO AI-GENERATED ART” launched by the Bulgarian artist Alexander Nanitchkov is gaining momentum on different networks, supported by many concerned artists. This movement originates from Nanitchkov’s will to protest against the plagiarism of some artists’ works on the specialized website ArtStation by AIs. This platform allows artists to give visibility to their works in order to sell them or to be spotted by potential employers.
In conclusion, while AI-generated art has enormous potential to revolutionize creative production processes across all sectors, legal tools for intellectual property are needed to protect artists from exploitation. This means that governments should consider enacting legislation that frames licensing agreements and payment structures for any artwork generated by AI systems not only for the visual arts, but also music, text, and other forms of expression. In doing so, we can ensure that creators, those unique individuals with unique voices expressing themselves through the medium of their choice, receive their due for their work and contributions that make our culture what it is today.
About this article
This article has been written by a student on the Grenoble Ecole de Management’s Advanced Masters in Digital Strategy Management. As part of a content creation assignment, students are given the task of writing articles based on their digital interests and disseminating the articles online. Articles are marked but we make minimal changes to the content. Thanks for reading! James Barisic, Programme Director, MS DSM.
