Artificial Intelligence and the Visual Arts: The Disappearance of the Artist?
By Sara Foster
Last month in an auction at Christie’s, a work by an unknown artist sold for three times the price of a Roy Lichtenstein and an Andy Warhol print. The work, titled Edmond de Belamy, from La Famille de Belamy (2018) was produced using Artificial Intelligence (Al) by the Paris-based collective Obvious. The development is nothing new, as artists such as Tom White, Mario Klingemann, and Robbie Barrat have been using AI as a medium for years. However, the staggering amount the work generated - $350,000 is significant compared to the original estimate by Christie’s which was a mere $7,000 to $10,000. The question is: why would anyone want to own a work of art created by AI? In an increasingly technological society, we are forced to examine what we want from art, what informs our relationship with it, and whether these have anything to do with the artist at all.
In 1967 French literary critic Roland Barthes published what would become perhaps his most celebrated and most controversial work, ‘The Death of the Author’. The essay was a response to the authority given to the artist throughout art history and the idea that a work’s singular meaning could be determined by the artist’s intention. Barthes concluded that all meaning could be found in the work itself, independent of its author. Two years later, Michel Foucault took up a similar position in a lecture entitled, ‘What is an Author?’ in which he declared contemporary writing had liberated itself from the necessity of ‘expression’ and could exist as an autonomous entity. The artist to Foucault was irrelevant. With the increasing prioritization of the text itself, the dehumanized artist became merely a machine, referred to by Foucault as the ‘author-function’. The individuality of the artist, the significance of his creative capacities, and the possibility of his genius, had been all but destroyed.
In the face of AI’s growing popularity in the visual arts, the role of the artist must again be examined. For works such as Edmond de Belamy, some argue that value does not derive from its aesthetic qualities or reference to artistic talent. Writing for Artsy, Ahmed Elgammal insists that “the art [produced by AI] is not in the outcome or final image, the art is in the process”. Elgammal considers that AI’s artistic production is simply a new form of conceptual art, where the idea behind the work is more important than the artwork itself. The process is as follows: the artist selects the input of images into an algorithm, and in the algorithm’s failure to make correct imitations of the pre-selected input it generates distorted images from which the artist chooses the final work. If it were not for this essential failure – if the algorithm was successful in imitating the input data – then Elgammal says the output “would not even be interesting as art”. It seems the compelling quality of the work is determined within the computer, and AI appears only to place limitations on the artist’s creativity. Why, then, would one rely on such a method? The appeal, simply put, is that it may help the artist and his audience better understand the aesthetic qualities of artworks and the processes by which they are produced. Computer programs can be tested and inspected in ways the actual artistic process cannot, as we have no access to the inner workings of the artist’s mind.
If the creative process of AI can be explained, it can then be copied, which raises the inevitable question of originality and the problem of attribution. Walter Benjamin’s famous essay, The Work of Art in the Age of Mechanical Reproduction (1936) first raised these issues in relation to the new technology of photography and film. Doubtless influencing the positions of writers such as Barthes and Foucault, Benjamin generally praised reproducible artworks for calling the superiority of the artist into question and making art more accessible to the masses. While the concern of reproduction is certainly integrated into AI-generated works, it is equally unclear where to locate the ‘artist’ in the creative process. If the process is thought to encompass the determination of input images, the operations of the algorithm, and the selection of the final image, then the work must be regarded as a collaboration between the person selecting the images and the one constructing the algorithm.
A more controversial position, of course, is that the ‘artist’ is the computer itself. On August 17th, Nature Morte in New Delhi – one of the largest contemporary commercial galleries in India – opened their exhibition entitled “Gradient Descent”, which exclusively showcased works generated by AI. One curator behind the show, Karthik Kalyanaraman, told The Verge that the works suggest computers may need to be taken seriously as creative agents, and that the ‘mysticism’ surrounding our accepted notion of creativity is what stops us from acknowledging them as such. “If a machine can make humanly surprising, stylistically new kinds of art,” he says, “I think it is foolish to say it’s not really creative because it doesn’t have consciousness”.
Kalyanaraman is right in one respect: the mysticism he refers to has always lent itself to art and the creative process. From the magical power of relics, to the divine nature given to artistic inspiration in the Renaissance, to our persistent reverence of artistic ‘geniuses’, there is a sense in which art, as the product of the artist’s mind, necessarily evades understanding. In the modern era, we have consistently attempted to overcome this obstacle and bring art down from its transcendental plane. Critics such as Benjamin encouraged this effort, commending modern technology for destroying the magical ‘aura’ of a work that derived from its uniqueness and authenticity.
Artificial intelligence, then, is our newest means to comprehend art by defying its inherent mystical qualities and questioning its originality. A position such as Kalyanaraman’s seems to represent the culmination of Foucault’s dehumanization of the artist. It insists that artistic creativity and expression is a mere program that can be reproduced and mimicked by a computer and its algorithms. Ultimately, we have much to lose with this perspective. The distinction between a human brain and a computer program is not a trivial one, and to decipher the process of creation behind an AI-generated artwork is not to understand the concept of human creativity. Our relationship to art, after all, is essentially human. Artistic expression is a distinctly human endeavour, an existential struggle to articulate and make sense of what is unknown or foreign to us. And so as the prominence of AI increases, the importance of the artist in our technological era is perhaps greater now than ever.
Barthes, Roland. ‘Death of the Author.’ In Image, Music, Text, ed. and trans. Stephen Heath. New York: Hill and Wang, 1978, pp. 142-148.
Benjamin, Walter. ‘The Work of Art in the Age of its Technological Reproducibility: Second Version.’  In The Work of Art in the Age of its Technological Reproducibility, and Other Writings on Media, eds Michael W. Jennings, Brigid Doherty, Thomas Y. Levin, 19-55. Cambridge, Mass,: London: Harvard University Press, 2008.
Elgammal, Ahmed. ‘What the Art World Is Failing to Grasp about Christie’s AI Portrait Coup.’ Artsy.com. https://www.artsy.net/article/artsy-editorial-art-failing-grasp-christies-ai-portrait-coup. (October 29, 2018).
Foucault, Michel. ‘What is an Author?’ Language, Counter-Memory, Practice: Selected Essays and Interviews by Michel Foucault, trans. by Donald Bouchard and Sherry Simon. Ithaca, NY: Cornell University Press, 1977, pp. 113-138.
Kugel, Peter. ‘Artificial Intelligence and Visual Art,’ Leonardo 14, no. 2 (1981): pp. 137-139
Vincent, James. ‘What Algorithmic Art Can Teach Us About Artificial Intelligence.’ Theverge.com. https://www.theverge.com/2018/8/21/17761424/ai-algorithm-art-machine-vision-perception-tom-white-treachery-imagenet(November 6, 2018).