The debate on the relationship between art and artificial intelligence (AI) has been heated for several years, but in 2022 it exploded in the media thanks to the appearance of text-to-image algorithms such as Midjourney, Dall-E, Crayon and Stable Diffusion, capable of generating images based on text instructions (prompts).
These AI algorithms are the result of the convergence of three technological “phenomena”: the growth of big data, the development of image recognition and generation algorithms, the development of algorithms for natural language processing and production.
Text-to-image algorithms combine these three components, as they receive a written verbal description of an image to be generated, understand the meaning of the text and then create an image that matches the described characteristics by finding similar features in the huge image archives they explored when trained.
While some see these algorithms as new tools to experiment with, as photography and digital graphics were when they first appeared, others see their use as troubling and a threat to the art industry.
Indeed, artificial intelligence is finding more and more use in numerous professional fields, such as communication, graphics, design, fashion, advertising and journalism, with a potentially significant impact on the work of artists and creatives.
Another concern often raised is the potential of AI generated images to be increasingly indistinguishable from artworks created directly by man.
AI algorithms are trained to recognize structures within real image catalogs, hence developing their own knowledge of the art world, classifying cultural heritage and moving between different artistic styles and movements. The set of images explored constitutes the imagery of artificial intelligence, the source of its historical-theoretical knowledge and its generative capacities.
The very use of archives consisting of existing works of art has raised questions about the infringement of the property of the material protected by copyright from which the algorithms are “inspired”, highlighting the current inadequacy of laws on this subject.
In December 2022, some artists in Italy created a crowdfunding campaign to cover the legal costs needed to push EU institutions to regulate how companies that produce algorithms for creating images collect their data. Concerns from the creative community led the European Parliament on May 11 to approve a new version of the AI Act, more seriously focused on the risks of generative AI.
Other artists have filed lawsuits against these companies for infringing the copyright of millions of artists to train their algorithms. However, some commentators think that these actions could lead to monopolies, favoring larger tech-companies with the financial capacity to procure large quantities of images.
The debate on the use of artificial intelligence in the art world also focuses on the ability of generative AI algorithms applied to art to produce results that can be considered artistic and on the aesthetic and cultural relevance of the images produced.
In August 2022, in an art contest in Colorado, one of the winning artworks was created using the Midjourney algorithm. The news immediately aroused the indignation of the web against the author of the digitally manipulated photography, Jason Allen.
However, many of those who have started using AI in the artistic field, such as Davide Quayola, Refik Anadol and Ian Cheng, were already recognized artists and inserted in the international art system, well before applying this new tool to their work.
The applications of artificial intelligence within art world institutions, such as large museums, foundations and public/private bodies, are mainly focused on the management of artistic heritage, rather than on the generation of new works of art. One such example is the use of facial recognition and detailed analysis techniques to attribute anonymous artwork.
Furthermore, artificial intelligence has been employed in experiments to aid in the restoration of artworks. Another intriguing application involves the identification of recurring visual elements within museum archives, which can then be compared with contemporary images. This comparative analysis provides a valuable insight into the consistent figural characteristics present across various time periods and artistic movements.
In this way, new horizons open up to the public and artists and the conception of art expands through the use of new technologies. However – as usual – to understand in which direction we are going we will have to wait some time for the whole art system to metabolize the change.