The gesture is singular: holding up a mirror to artificial intelligence. Placing it at the heart of artistic production to better reveal its mysteries. More than a tool, AI is here both an object of reflection and a stakeholder in creation, the majority of the creations exhibited having been created by artists “on” and “with” AI, in a form of collaboration between human and machine. A dialogue between the artist and the scientist. At the intersection of art and science.
“This exhibition focuses specifically on how algorithms and AI models transform our experience of images, both still and moving, and the role they play in culture,” writes Antonio Somaini, chief curator, in the introduction to the exhibition catalogue. The aim is to uncover the “discreet operations” and “invisible processes” behind artificial imagery. To probe the black box of AI.
New technology neophytes, don’t panic! The frescoes and diagrams by researcher Kate Crawford and designer Vladan Joler, XXL infographics covering the walls, immediately retrace the history of the techniques and thought systems behind AI, from the Renaissance to the present day. These landmarks provide much-needed context, offering an overview of this complex interweaving.
Along the way, the viewer is immersed in “latent spaces,” these artificial neural networks capable of producing images endlessly from combinations of numbers and codes. Welcome to the matrix! In this desire to make the work of algorithms tangible, the installation by American Trevor Paglen scans the faces of participants in real time, retransmitted onto a giant screen and then associated with a database. A personalized labeling, in short.
Showing the invisible
Surprisingly, AI is anything but immaterial. It’s essential to deconstruct this myth, according to Antonio Somaini. “AI models rely on extremely energy-intensive computational processes and the massive extraction of non-renewable natural resources.” Copper, lithium, cobalt, water, minerals. The devastating extraction of these raw materials and the desolate landscapes they leave behind are at the heart of the work of French-Swiss artist Julian Charrière.
In “Buried Sunshines Burn” (2023), he transfigures Californian oil drilling fields into heliogravures (photographic prints on metal). The golden and silver reflections of the materials highlight, as in a reflective mirror, the soil contaminated by black gold. This ecological pollution is also visible in the sculpture “Metamorphism” by the same artist: an agglomeration of electronic components and earth that warns of the toxic waste generated by AI.
This is one of the strengths of this exhibition: to restore artificial intelligence to its harshest materiality. And to draw attention to its flawed ethics. It focuses in particular on the deplorable living conditions of AI’s little hands. “All over the world, millions of people perform underpaid micro-tasks, without any social protection, to organize training data sets and supervise content generation,” recalls Antonio Somiani.
German director Hito Steyerl takes us to a refugee camp in Kurdistan to meet the click workers. Tasked with labeling the data needed for artificial intelligence models, they share their daily lives in a 13-minute video. The height of cynicism is that by endlessly indexing identified objects, they contribute to the training of drones that could be used for military purposes against their own population.
By portraying these “ghost workers,” artist Agnieszka Kurant gives them a composite face. An identity. A semblance of dignity. For its part, the installation by the Meta Office collective, in collaboration with these digital stokers, reveals their workplace to illustrate the precariousness of this new working class exploited outside of any legal framework. Behind the screen, the nightmare of the human condition.
Ghosts and hallucinations
Invisibility remains. As we know, AI is not neutral and is not immune to the biases of domination propagated despite themselves by its designers. This is the case with the facial recognition system. To denounce the stereotypical and colonial vision of its algorithms, New York-based artist Nouf Aljowaysir chose to erase the silhouettes of humans and animals from thousands of photographs from online collections relating to her regions of origin, Iraq and Saudi Arabia.
The technique is reminiscent of the empty chair: giving a metaphorical presence to those forgotten by history by highlighting their absence. “Through this gesture, she gives them a paradoxical spectral presence. It is thus as much a matter of escaping orientalizing clichés as of exacerbating to the extreme a lack of visibility,” writes researcher Ada Ackerman. Further on, a constellation of ghostly subjects invite us “from their gaping hole, to renew our gaze on them.”
Other pieces reveal dystopian worlds. With “Fourth Memory,” an installation made up of images, videos, and sculptures generated by AI, French-Canadian Grégory Chatonsky immerses us in a chilling world worthy of the hit series Black Mirror. The artist says he fed artificial intelligence with his personal data—his voice, his photos—in order to create works after his death, which he describes as “posthumous works during his lifetime.”
The approach echoes the work of Trevor Paglen, who, with “Even the Dead Are Not Safe,” creates composite portraits of celebrities, such as Simone de Beauvoir, from images from the past selected by a facial recognition program. The result is a portrait that is larger than life. It is disturbing to think that this photograph is merely an illusion. Paglen disrupts our relationship with time, with reality. He blurs the line between the world of the dead and that of the living.
These radical artistic experiments, accompanied by an arsenal of critical reflections, are the essence of the exhibition. They reveal the infinite possibilities of AI, while pushing the technology to its limits, to the point of causing it to glitch, making it produce mind-boggling texts and visuals. “Hallucinations,” as the jargon goes. A certain poetry emanates from them. And the intuition of witnessing the birth of a new grammar of the image.
“The World According to AI” is on view at Jeu de Paume, in Paris, until September 21, 2025.
