Art is an imitation of life, Plato had argued in his theory of mimesis. What the Greek philosopher said millennia ago, economists are now trying to prove with the help of Artificial Intelligence. Researchers from the Sorbonne School of Economics, University of Bristol and the University of Toronto have studied 6,30,000 paintings from 1400 CE onward to uncover how visual art reflects the socioeconomic context in which it was created. To do this, they trained a learning algorithm to predict nine basic emotions — is this a wide enough range when it comes to something as diverse and evolving as human emotion? — conveyed in each painting. The researchers argue that the paintings can serve as valuable sources of historical information about societal moods and structural shifts, especially in periods where traditional economic and social data are scarce.
Several questions arise about this line of enquiry. For example, what kind of a worldview would AI paint on the basis of a dataset of works by predominantly White, Western artists? After all, it is well-documented that artists of colour have been overlooked by art historians, collectors and conservators. Their views of the economic conditions of a shared timescape are likely to be markedly different — for instance, when the American artist, Winslow Homer, was producing stunning seascapes, his contemporary, Bill Traylor, a slave most of his life, was expressing on canvas the fear, rage, and torment that African-Americans had to outwardly suppress in a segregated society. That is not all. Conflating a period of economic bounty with an ambience of joie de vivre can be erroneous. This is because art often expresses inner demons; the sadness or angst — emotions that AI can be taught to recognise — may not always be concomitant to the joyous world outside — consider the despondency of Vincent van Gogh’s portraiture at a time of relative bounty in Paris.
However, this is not to say that art and material conditions, a point that the economists and AI have underlined, are unrelated. There would be no Botticelli without the patronage of the Medicis; in classical Greece and Renaissance Italy, rivalries between competing patrons and towns led to technical and stylistic innovations and the rise of different art schools. Today, Donald Trump’s disdain for his art is getting Rigoberto A. González more attention than he would have received otherwise. Not just economic but social and ecological conditions also impact art and its perception. A porcelain urinal purchased by Marcel Duchamp would not be considered art had it not come in the midst of an industrial capitalist boom; the works of J.M.W. Turner have found a new lease of life in the 21st century for being some of the earliest chronicles of the changing climate.
A study of the relationship between art and economics mediated by technology can be productive and open up new vistas. But where AI would trip is in deciphering art’s emotional content. AI could read nine emotions in over 6,30,000 paintings; humans have detected more than six emotions in the enigmatic smile of one painting alone: the Mona Lisa. Algorithms may help trace material realities
but one requires the human mind’s eye to grasp the layered contradictions, ambiguities and depths that give art its true meaning.
