Amélie Boutinot
Associate Professor in Management Science, PhD, HDR
University of Strasbourg
Helene Delacour (photo)*
Professor in Management Science
University of Lorraine
*Faculty member of the Business Science Institute.
Article originally published on The Conversation France.
On October 25, 2018, Christie's, the famous auction house, offered for sale the portrait of Edmond de Belamy "painted" by an artificial intelligence (AI) program developed by the French collective Obvious. The sale was a success, as the portrait of this fictional young man, looking like a 19th century notable with a blurred face, was sold for $432,500, whereas it had been estimated at $7,000. Despite the success of this sale, a question arises: can artificial intelligence be considered an artist?
A four-point definition
According to the American sociologist Becker (1982), every artist is included in a "world of art" which gathers the artists who produce a work, but also all the actors and the trades participating directly or indirectly in their creation: for example, the painters cannot practice their art without brush or canvas and sell without gallery, nor painting salon. All these actors like critics, collectors, curators, art historians, etc. interact in a regular way around the "art world".
The notion of "art world" implies that an artist seeks recognition not from a single audience (usually peers) but from a plurality of audiences directly or indirectly connected to a work of art, each with its own frame of reference and values. In this respect, Bowness reveals that an artist builds his recognition with four concentric circles: with his peers, then with art or cultural critics, then with the market, and finally with the general public, whether connoisseur or novice in the discipline concerned. This is quantitatively the most important public, but with whom the artist has very little contact.
In creative activities, such as painting or singing, recognition by a large public can be perceived as a driving force for an artist, but must be established over the long term. Indeed, too rapid access to success with the general public does not allow for the establishment of legitimate recognition in the eyes of artists; such success would be the mark of a lack of creation.
Other authors have developed this particular relationship with recognition by a large public as necessary but very complex.
Like Baxandall (1985) who recalled the uncertainty of buyers with regard to Renaissance painters, where it is only during the transaction that they will be able to judge the quality of the commissioned works, Caves (2000) underlines that evaluation by price functions with difficulty in artistic activities.
In the world of contemporary art, to detect the aesthetic talent of an artist, a classification by sale price of works (notably in auction rooms such as Sotheby's or Christie's) makes it possible to identify the artists in vogue, on whom clients invest].
In the same way, the Kunst Compass is a kind of artist recognition scale that proposes an objective measurement of the degree of recognition of an artist through different weighted criteria. At the same time, art fairs tell collectors which artists to invest in, and mediators, as cultural intermediaries and/or critics, guide potential customers to the artist in vogue with their articles.
AI, an artist like any other?
Starting from this observation, if the AI cannot yet be considered as an artist in its own right, in the traditional sense of the term, it nevertheless shares certain characteristics. First of all, it should be underlined that the AI having realized the portrait of Edmond de Belamy has, even without real consciousness, nor intention proper to human artists, created something that did not exist before.
As Hugo Caselles-Dupré, one of the founders of the Obvious collective, points out, "artificial intelligence can be creative. [We force it to create a new visual". Moreover, the painting in question is rooted in an already known genre, that of the portrait, whose codes have been studied by the AI program on 15,000 paintings dating back to the Middle Ages. It can, therefore, be categorized and labeled, whether by artists, buyers, collectors, or critics. This same painting was also sold at Christie's, a world-renowned auction house - and thus integrated into an "art world".
At the same time, another portrait designed by this AI was also sold for nearly 10,000 euros in 2018 to an avant-garde collector, another influential audience integrated into the "art world". Beyond this membership in the "art world," this collection of portraits demonstrates that the AI in question has generated not one but several "paintings." The portrait of Edmond de Belamy is indeed part of a set of "works", with eleven portraits of this fictional family, the Belamy, from the same algorithm.
Moreover, the portrait is signed with the equation of the program that generated it, thus claiming the paternity of the conception and realization. Finally, the portrait is claimed as belonging to the "ganism" movement (GAN for generative adversarial networks, i.e. the technology used), of artistic movements.
A complex field of reflection
However, the second and third characteristics linked to the quest for recognition open a more complex field of reflection. Indeed, the term quest is not adapted in itself for an AI, based for the moment on a human-machine interaction, where it is the human (here the Obvious collective) who can seek a form of recognition.
Concerning the quest for long-term recognition, the Obvious collective, in its will to democratize, has not expressed a wish for long-term recognition (thus working for its posterity, for that of the AI), nor for short-term recognition. If the works created by AIs can generate a certain media buzz, this is not a guarantee of posterity.
In the case that interests us, it is also the man who stages and promotes the achievements of the AI, and has the intention proper to the artists.
Finally, concerning the characteristic of the uncertainty proper to the detection of the artists, the stake for an AI is not so much to valorize by its achievements, as to evaluate the technical and financial resources to realize the digitally created works. Indeed, the manufacturing of a painting generated by such a computing power is expensive. As the collective has no financial support as such, belonging to the "art world" appears necessary to find financial resources.
The AI is therefore not an artist like the others. But this does not mean that AI is not an artist, if we consider that the boundaries of this term can evolve. The advances of GAN will certainly allow to solve the problem, by granting a creative capacity to AIs. Tomorrow, a new profile of artist could well emerge, with more digital characteristics. AI would no longer be considered as a super brush or a technical tool for human use, but as an entity generating artistic creativity. The question will then be: what will be the new profiles of artists?
Before arriving at that, the actors of the "worlds of art" will have to answer other questions, of a legal nature. Indeed, the authorship of the work, which is already posed vis-à-vis the program or the collective, extends to another artist, Robbie Barrat, having deposited the source code in open source in early 2018... And who asks to be included in the reflections.
Article translated from French with https://www.deepl.com/translator
Read also...
Hélène Delacour's articles on The Conversation France.
Hélène Delacour's articles & books via CAIRN.Info.