Two people dressed in black are kneeling on the floor, so still that they must surely be in pain. If they are grimacing, there would be no way to know â their features are obscured by oversized, smooth gold masks, as though they have buried their faces in half an Easter egg.
Their stillness makes them seem like sculptures, and only by checking for the subtle rise and fall of their chests can you confirm they are indeed human. Which is fitting, really â because they arenât actually human, at least not totally. Theyâre human-machine hybrids, âIdiomsâ, created by French artist Pierre Huyghe for his largest ever exhibition, Liminal, at the Punta della Dogana in Venice.
Idioms are wandering the exhibition for its run between March and November. Sensors in their masks monitor the rooms they sit in and visitors they encounter, and artificial intelligence will gradually convert this information into a brand new language. Slowly, for example, the Idiomsâ masks will come up with the words for âdoorâ or âhumansâ or âwritingâ â building a dictionary until they will even be able to communicate with one another. Every day, their knowledge will accumulate; Huyghe wonders what they might be able to say in 20 yearsâ time.
On a crisp March day, shortly before the exhibition opens to the public, two Idioms kneel in a darkened room opposite a large black box suspended from the ceiling â this is a âself-generating instrumentâ (also loaded with environmental sensors), producing ambient music and crisscrossing beams of light. In response to the artwork in front of them, the Idioms appear to have only generated a few syllables, repeated intermittently over and over again as the LED screens on their foreheads glow gold. Their words are a hissing whisper. It sounds a lot like, âWhatâs this?â
Liminal, with Huygheâs Portal, a sensory antenna and transmitter, in the centre Photograph: Ola Rindal/Palazzo Grassi, Pinault Collection
Itâs a fair question to ask. The dilemma facing any artist who tries to tackle a subject as paradigm-changing and era-defining as artificial intelligence is that the real magic is often happening on some hard drives behind the scenes. While there is a blinking server on show at Liminal, Huyghe himself conceded at a press conference three days before opening that it might be hard for a casual visitor to understand that the language coming from the Idiomsâ masks is AI-generated; he worried that visitors would assume that the people wearing the masks are the ones whispering.
For contemporary artists, there is a clear pressure to tackle and engage with the buzzy technology that has rapidly disrupted everything from homework to journalism since ChatGPTâs debut in 2022.
Like Huyghe, creatives from German film-maker Hito Steyerl to British conceptualist Gillian Wearing have used AI to make or enhance their art. Shortly after Liminalâs first run closes, an ostensibly âfully AI-drivenâ multimedia exhibition of French artist Philippe Parrenoâs historical works will open at Haus der Kunst in Munich.
Whether artists are using the technology in an interesting and challenging way or simply hoping to hop on the hype bandwagon is not always easy to discern. From a preliminary press release of the Munich show, itâs unclear exactly which elements of Parrenoâs exhibition will be artificially intelligent, and itâs easy to see how AI could cynically be slapped on to an exhibition like an Instagram filter, a shiny veneer that makes old work seem new.
AI is already all around us, autocompleting our emails, suggesting a new show to watch on Netflix, and reading the weather forecast in the voice of Amazonâs Alexa. In recent years, chatbots have revolutionised writing â responding to prompts to write cover letters, code, plays, poems, and essays â while text-to-image models such as DALL.E and Midjourney allow anyone to create âartâ by typing in a few words.
Echoes of the Earth: Living Archive by Refik Anadol at the Serpentine Gallery, London. Photograph: Guy Bell/REX/Shutterstock
But as the technology becomes more prominent in our everyday lives, artistsâ use of AI risks feeling trite. Crowds have allegedly been âtransfixed for an hour or moreâ by Turkish artist Refik Anadolâs âlive paintingsâ currently being displayed at the Serpentine Gallery in London. AI was fed imagery of rainforests and coral reefs to generate Anadolâs exhibition, Echoes of the Earth: Living Archive, which features immersive âartificial realitiesâ that visitors can wander through. While crowds may be transfixed, critics have said that Anadolâs previous AI-generated work is over-hyped.
âThe whole thing looks like a massive techno lava lamp,â New York Magazineâs Jerry Saltz wrote of Anadolâs Unsupervised, a 24ft screen that used AI to continuously generate images at the Museum of Modern Art between 2022 and 2023. Saltz found the work to be pointless and mediocre â good at entertaining you briefly but ultimately ânot disturbing anything inside youâ. In short, he felt the work had nothing to say.
Saltz argued that âif AI is to create meaningful art, it will have to provide its own vision and vocabularyâ. On a literal level, this is exactly what Huygheâs Idioms are doing. Watching them is oddly mesmerising â as a viewer, it is interesting to be confronted not with a finite state of âartificial intelligenceâ, but an ongoing process of âartificial learningâ.
Here, Huygheâs use of AI takes the art out of the artistâs control, which is exciting â not least because of the possibility that things could go wrong. The Idioms could fail to produce a language or produce one that is discordant and offensive to our ears. They could be unduly influenced by rowdy exhibition-goers or rebel in some way, repeating the same words over and over again.
skip past newsletter promotion
Your weekly art world round-up, sketching out all the biggest stories, scandals and exhibitions
Privacy Notice: Newsletters may contain info about charities, online ads, and content funded by outside parties. For more information see our Privacy Policy. We use Google reCaptcha to protect our website and the Google Privacy Policy and Terms of Service apply.
after newsletter promotion
It would undoubtedly be fascinating to return day after day and see how the Idioms have responded to the art around them. As Huyghe intended, these strange masked beings provoke questions about the relationship between the human and the non-human (even if my first thought was, âI bet their knees hurt from all that kneelingâ).
Endless editing process ⦠a still from Camata by Pierre Huyghe. Photograph: Pierre Huyghe/Courtesy the artist and Galerie Chantal Crousel, Marian Goodman Gallery, Hauser & Wirth, Esther Schipper, and TARO NASU by SIAE 2023
Less thought-provoking is the use of AI in his work Camata. Robotic limbs surround a skeleton in one of the worldâs driest deserts, performing a mysterious ritual. Though the footage is not live, the film is edited in real time, with artificially intelligent âeditorsâ gathering data from a large brass telephone-pole-like sensor near the opening of the exhibition. This sensor monitors everything from the number of guests in the gallery to the weather outside, and the Camata footage is edited accordingly.
Yet curator Anne Stenne clarifies that this isnât a simple case of âxâ leading to âyâ â for example, if only one person was in the exhibition, it wouldnât be the case that the AI editors would automatically, say, choose footage shot at night. This means that while the endless editing process is fascinating â you could, after all, sit there for the exhibitionâs entire run and never see the same sequence twice â itâs hard to understand as a layperson why AI was a necessary element. Would the work be any different if the editing was randomly generated? As a casual viewer, itâs very hard to know.
Indeed, those who attend these exhibitions simply have to trust that something fantastical is occurring behind the scenes. While Huygheâs sensors are visible throughout the exhibition, the artist is unwilling to share the details of the program that processes this information and exactly how it runs. A representative says, âPierre doesnât want to focus on the technical parameters of his works. He wants to concentrate on the visitorâs experience.â Audiences may find this troubling in a world where companies have been found to be using âpseudo-AIâ that is actually run by hidden humans behind the scenes.
AI art works best when it does something that the artist alone could not, as is the case with Huygheâs self-generating language. Anything else risks feeling at best gimmicky and at worst pointless. Regardless, the AI trend will continue to sweep galleries, and soon enough the tool will be commonplace enough that questioning it will be like questioning a pen or a pencil.
In the 1960s, âcomputer artâ swept the globe, with exhibits from London to Stuttgart to Zagreb to Las Vegas. One contemporary writer said âperhaps a computer will never produce a painting all by itselfâ, and noted with caution that âat least one expert thinks such art represents a genuine new art formâ. One day, undoubtedly, discussions of AIâs place in art will sound this archaic.
Liminal by Pierre Huyghe is at Punda della Dogana, Venice, until 24 November