Artificial Intelligence in Global Art Auctions Can works created by AI be considered works of art and who owns copyright?


Nowadays, artificial intelligence (hereinafter AI) is conquering almost all spheres of life, and the higher art industry is no exception. The works created by AI end up in global art auctions, stand alongside the works of great artists, and compete in the fight of imagination and creativity. Indeed, the creative potential of the machine is still at its early stages, but we already have examples shaking the imagination of utopians. Moreover, AI is also used to improve the functioning of the auctions themselves by expanding the audience of potential art users and by increasing the number of people involved in art procurement/sales processes, including the lower classes. And while the computer is still under human control, we are in the face of revolutionary “machine learning.” 

Portrait of Edmond de Belamy, GANs Algorithm, Inkjet printed on Canvas, 70x70 cm

Portrait of Edmond de Belamy, GANs Algorithm, Inkjet printed on Canvas, 70×70 cm, source:

Application of new technologies in auction activities 

The oldest auction house, Sotheby’s, founded in 1744 in London by Samuel Baker, is among the most innovative. Traditional methods for evaluating works are complemented by innovative ideas of the millennials. In 2018, a technological start-up, Thread Genius, was installed at the auction, becoming part of the digital strategy. Based on visual recognition, today’s extremely popular algorithm selects works of art suitable for a potential buyer, just like those in Spotify or Netflix, which regularly provide recommendations on what to listen to or watch. At the same time, the Thread Genius program, based on Sotheby’s auction database, can anticipate future material changes in the value of the work or even predict trends in art. True, competent opinions of art historians and critics remain a privilege, and the higher class is generally reluctant to rely on the “taste” of AI, which is merely based on programmatically defined guidelines for the evaluation of works, such as formal elements. Buyers who perceive art as a valuable investment are usually aware of their preferences and the value of the work they buy, while personal connection and the emotional experience of the work are important measures of the value of art. As a result, artificial intelligence is rather an auxiliary tool for accelerating the work of art experts and their long-standing research, expanding the range of potential non-selective buyers, and relatively facilitating the acquisition and sale of works of art.

The most expensive works created by AI

The most expensive works created by AI Christie’s, the world’s largest auction house, founded in London in 1766, regularly monitors the impact of technology on the art market and actively responds to its changes. With over 350 biddings held each year, the auction has also provided a public platform to present the first work created by AI. The successful debutants were a trio of artists from France, Obvious Art, and their work Edmond de Belamy, from La Famille de Belamy, auctioned for nearly half a million US dollars (432,500) exceeding the original price more than 40 times. The artistic team of Obvious Art consists of Hugo Caselles-Dupre, Pierre Fautrel, and Gauthier Vernier, who widely use mathematical algorithms in their work, question the concept of creativity machine, dressed in a dark cloak with a white collar, suggests that the person is from the church class. The modern-looking portrait was actually created using databases of as many as 15 thousand portraits painted between the 14th and 20th centuries. The GAN (Generative Adversarial Network) algorithm was used to generate a unique parallel computer reality based on the formal features of existing images. Interestingly, GAN not only generates images and mimics patterns of human behavior, but also reveals an autonomous ability to develop creative processes and model the course of art history. The question of who owns the copyright of this work is resolved in this case by the initials written in the corner of the painting. They are not human, though, but part of the machine algorithm code that created this portrait. Indeed, Obvious Art are not the first to take advantage of the creative opportunities offered by AI. Here, Mario Klingemann, a German scientist and pioneer of the AI artistic movement, has been working on algorithm-based art for 25 years and has generated several hundred, he says, low artistic value, primitive and repetitive pictures. As a result, he emphasizes the creative process when applying computer technology, placing it above the final and static result, even though the latter has quite a strong composition and emotional influence.

The artist’s most famous work, Memories of Passerby I, despite becoming a sensation of Sotheby’s auction in 2019, reached a much more modest amount of £40,000.

The installation consists of two screens, an antique cabinet with a beautifully hidden AI “brain”, and a comfortable chair for the viewer to sit down and immerse oneself in a continuous automated artistic performance. By creating a cozy work form and a contemplative space, the artist, like the creators of Obvious Art, seeks to democratize the AI and destroy myths about its alleged threat. The changing images on screens are also known as “neurography,” a continuous sequence of images generated by artificial neuronal networks. The work was based on the GAN algorithm and enormous Western European art databases from the 17th to the 19th centuries. By interpreting input data in a variety of ways, the computer continuously generates fictitious portraits of men and women, and the freely available online data allows the program to constantly improve. Mr. Klingemann uses AI as a tool, an instrument to create. Like with brush strokes, the artist “paints” by using a computer keyboard and managing neural networks, codes, and algorithms. This collaboration between the artist and the computer is like a manifestation of double intelligence, human and machine, in the work. Although at first glance, Memories of Passerby I seems to be operating autonomously, it needs to be turned on for the program to create. Man becomes an inevitable and almost essential part of this artistic chain. The latter human “power” to turn the computer on/ off gives hope that in the world of the future, control will remain in people’s hands.

Reasons for Obvious Art’s success 

The creative trio of Obvious Art are not the only ones to have benefited from the creative opportunities provided by the GAN algorithm. Rutgers University in New Jersey, USA, for instance, has its own Laboratory of Art and Artificial Intelligence. Many programmer-artists publish their works on social media and exhibit in galleries. But why was it the work of Obvious Art that was valued at a record amount thus stirring the imagination of technology and art speculators and paving the way to the utopian future of art? The success of the Edmond de Belamy portrait seems to rest on the Institutional Art Theory, where the value of the work is determined by the authorities and institutions of the art world. In this case, Christie’s, the world’s largest auction, “baptized” the latter piece of art created by AI and placed it next to the masterpieces of Andy Warhol and Roy Lichtenstein in 2018. However, it is also worth bearing in mind the contribution of the artists themselves to the promotion of the work. Using a widely available program, the Parisian trio presented the increasing influence of AI in the creative process, raised important questions not previously asked in the public arena of art – can works created by AI be considered works of art and who owns copyright?

read also JSC Berlin. Photo by Robert Hamacher. Courtesy of Julia Stoschek Collection.

7 Collectors to Follow Online Each of them collects with a different goal in mind, but they can all help keep us in the loop about the newest art trends

Aleksandra Mainka-Pawłowska Aug 18, 2020

Collectors have always played a significant role in the art world – they are trendsetters who actively shape the art market and influence it. But some of them are also real celebrities whose glamorous lives and eccentric behaviour resemble those of movie stars. Following their social media accounts can give us an insight behind the scenes of the art world, filled with champagne and great art.

Following our list of seven artists to follow online, we present a list of seven collectors active on social media. Each of them collects with a different goal in mind, but they can all help keep us in the loop about the newest art trends.

On the one hand, we could consider a person to be the author because the creation of a work was the result of many actions he initiated, such as creating an algorithm, entering data, and ultimately launching a program. On the other hand, thanks to the autonomous processes taking place inside the computer, we can see works with new aesthetics. Other work created by Obvious Art was also among the sales of art auctions. But the consortium of private art collectors only reached 12 thousand US dollars and confirmed society’s skepticism aimed at art of inhuman origin. One of AI’s works, Katsuwaka of the Dawn Lagoon, from the 22-picture series Electric Dreams of Ukiya, appeared at Sotheby’s auction in 2019. Using the aforementioned GAN algorithm, the artists initiated a series of works in which AI processed original Japanese woodblock prints created in 1780-1880 and produced new original images with initials – stamps – that became a reference to the artistic movement of ukiyo-e. The traditional genre of the Japanese press, which flourished in the 17th – 19th centuries, and the emergence of electricity, served as a starting point for talking about the ongoing entrenchment of AI in various spheres of human life today.

Copyright Issues 

Naturally, where the creative human mind and artificial intelligence meet, there is a fundamental question of the authorship of art, based on the concept of art in a broader sense. Say, if we consider art as something that responds to the subjective world of an individual, creation as a manifestation of an autonomous will, AI will not be able to fulfill it at least for the time being.

However, if we call art intellectual activity and the pursuit of an aesthetic result, the computer could be considered the author of works. The aspect of the artist’s perception also becomes relevant. One of them is that the creator is the one who creates the images, while art is the visual expression of the form. In the traditional conception of art, a formalistic approach is beneficial to the computer because it, too, can generate new creative images. It’s different if we consider an artist to be a creator who carries a conscious message to the public with his work. A person, using a computer and its capabilities, seeks to answer the questions asked. Such peaceful collaboration between man and machine suggests that the computer is rather a means of expression, a medium that helps to reveal the author’s ideas.

read also

Art & Law: Copyright or Licence? Important Differences

Oskar Roesler Mar 29, 2021

In this article we will discuss issues related to the transfer of economic copyright and licence, we will try to present a simple way of distinguishing between these two legal constructs, and present recommendations regarding the selection of the most appropriate solution for various types of contracts governing the trade in works of art.

In this new series (appearing monthly) we discuss the basic legal aspects of working as an artist, freelancer or business owner in the culture sector. It has been drafted exclusively for Contemporary Lynx by Oskar Roesler.


The emergence of a new genre of art 

Sotheby’s and Christie’s are not the only auctions to sell works created by AI. In 2016, Google organized a charity auction Gray Area Foundation for the Arts in San Francisco. It resulted in the sale of 29 AI-created paintings, which in total reached USD 100 000. The auction featured works generated using the open-source Google program DeepDream, which was developed by Google engineer Alexander Mordvintsev in 2015. The program consists of artificial neural networks based on the functioning of the biological nervous system. Combined computer neurons not only allow the processing of enormous amounts of data, but also the generation of new original images. The style of the paintings created by DI, which is called Inceptionism, is characterized by surreal motifs, and the images often resemble psychedelic fractals and phantasmagoric compositions of plants and animals. According to the program’s developers, Inceptionism allows for a deeper look at how your computer learns and copes with the set tasks, such as transforming specific objects into new images. Repetitive tendencies have been observed, when at first glance a primitive image is transformed into complex structures – rocks and trees turn into strings of buildings, horizons into exotic panoramas, while leaves turn into birds and insects. The latter transformations happen due to pareidolia, where random images are perceived by the computer as specific objects. Interestingly, the concept of Inceptionism originated from the widely known film Inception which is also dominated by surreal dream images. Artwork created by AI, which is astonishing at first glance, is merely the result of the evolution of science. At all times, art and scientific inventions have gone hand in hand – artists have used geometric shapes to show perspective, and the development of chemical science has contributed to the extraction of oil paint pigments. Today, the tools provided by the computer are becoming media that allow an increasingly wider audience to express themselves creatively and mentally. On the other hand, computer-generated images lose an aspect of the creative ritual. After all, the synthesis of decisions made and emotions experienced is an indisputably important part of a work of art, and the value of the work is created not only by the result but also by the peripeteias of the creative process. Works created by AI evoke ambiguous feelings. The images generated by the machine seem to lack originality and authenticity, while on the other hand, it is learning and developing surprisingly rapidly.

Patchlab 2018, vernisage, photo: Michał Ramus
HUMAN STUDY #1. 5 Robots Named Paul, Patrick Tresset, Patchlab Digital Art Festival 2018, vernisage, photo: Michał Ramus

Return to the homepage

About The Author


Past LYNX Collaborator

This might interest you