A discussion with Pierre Huyghe about ‘Variants,’ a sprawling new outdoor work near Oslo, in which the natural world and artificial intelligence restlessly entwine
For more than three decades, Pierre Huyghe has elided contemporary categories of art-making beyond the bounds of common syntax. Sculpture, performance, video, film, sound and land art become, in his hands, unpredictably permeable and unstable—a retort to what he once called ‘hysteric objects,’ a pursuit instead of a kind of art ‘somehow indifferent’ to the viewer. A reclining female nude concrete figure accommodates a hive of living bees in its head (‘Untitled (Liegender Frauenakt),’ 2012); an upstate New York community parade and summer gathering becomes a loosely scripted performance piece and film about human ritual and civic rites (‘Streamside Day Follies,’ 2003); glass-tanked aquatic environments inhabited by crabs and other sea life become what Huyghe has described as ‘non-illusionistic fiction,’ sculptural arenas for living elements to enact uncontrolled narratives within the parameters of constructed conditions (‘Zoodram’ works, 2009–13).
Two years ago, Huyghe was invited to consider creating an outdoor work within the environs of the Kistefos Museum’s sculpture park in Jevnaker, Norway, north of Oslo, on lush, rolling woodland near the Randselva River, once the site of a paper pulp mill. After some exploration of the area, he chose a portion of the museum’s site that had never before been a location of artwork, a small island that has now become the environment for Huyghe’s largest work to date, ‘Variants’—a word that has recently taken on new and ominous meaning. The work, a commingling of sculptural and digital elements, utilizes the land and the water that sometimes rises over it as the foundations for a loop in which nature and artificial intelligence systems feed into each other, creating what, over time, could come to function like a hacked ecosystem.
Huyghe sat down recently for a virtual discussion about the work and the ways in which ‘Variants’ has allowed him to delve deeper into several themes he has been exploring over the last decade. These are edited and condensed excerpts of the conversation.
Randy Kennedy: What was it about the wooded, watery territory at Kistefos—a piece of land that wasn’t previously part of the sculpture park—that interested you for creating a work?
Pierre Huyghe: I was looking for a milieu that would modify over time, as you can imagine from the works I have done before. And I realized that there was an island along the river below the old pulp mill factory. The cherry on the cake is that, occasionally, a dam opens and the water flows over a large part of the island. The opening of the dam is unpredictable, so randomly things appear or disappear. Scattered and entangled on the island are trash remains either carried by the river or left over from the factory—parts of machines, fragments of boats, bricks.
RK: And did the fact that the area is at first glance Edenic, idyllic, but when you look close it’s not quite, it’s corrupted in a sense, interest you?
PH: Yes. At first, it appears as a wild forest on an island. You notice the effect of the current within it, the way over years the flow has forced the trees to bend unnaturally. And then you notice the remains of industrial elements all over. The island being shaped by human controlled flooding was an interesting starting point for a speculation.
‘I wanted the actual to host a possibility of itself. The island and what this island could be in an alternate reality, a second milieu.’—Pierre Huyghe
RK: In other places where you’ve worked outdoors, like documenta 13 in 2012, or when you did the filming in New York City in 2005 inside Wollman Rink, the skating rink, for ‘A Journey That Wasn’t,’ those were, to a degree, human-structured environments. But in Kistefos you seem to be getting as close as possible to ‘unfettered nature.’ Did you want to take elements like machine learning and meld a fictional reality out of nature that’s not been touched tremendously by humans?
PH: Yes and no. I was looking for an entity, constituted of different milieux, that could react, generate, and I was looking for different intelligences at play. One milieu was given: the natural. I wanted the actual to host a possibility of itself. The island and what this island could be in an alternate reality, a second milieu. We scanned the whole island. It became the environment of a live simulation, in which mutations of elements, existing on the actual island, are generated by machine learning, an artificial neural network. The entire island is covered by real time sensors that capture bio and geo-chemical activity that modify these digital mutations. Some mutations exit the simulated environment and manifest physically on the actual at the exact location they were in the simulation, where they get corrupted, disintegrate, and change the actual island. I needed a space where all this could happen, somewhere that offered more than the conventional types of an art site, museum, structured sculpture park, pristine places with restrictions and constraints.
RK: Right. And those kinds of places also impose their own narrative.
PH: Exactly, and this place could offer alternative ones. But it is an illusion of untouched nature, as every two feet, there’s a trace of human presence. It’s just more diffused. And yet it appears more untouched than many places where I’ve worked—more trees, more biological activity.
RK: In works like ‘The Host and the Cloud’ and ‘A Journey That Wasn’t,’ there are people involved, sometimes many, or at least people are referred to in those works. Even in documenta, which was pretty un-people, there was the figure of the dog keeper. But it seems that in this work, and maybe others recently, that the pieces are moving in a direction in which humans have become absent from the scene. Is that purposeful?
PH: Right, recently the human has become more residual, not the main protagonist. I’ve tried to find other types of interactions or relations that are not human-centric. It’s less anthropocentric. In ‘Variants’ the permeability of an artificial neural network generating mutations and the bioactivity modifying it puts non-human types of intelligence into the position of being the main characters. There is a team of humans engineering this, of course, but it’s just that they are not involved as characters.
RK: I know that you’ve said a few times in different ways that you wanted to stop exhibiting things to people and you wanted instead to exhibit people to things, to a degree.
PH: That’s right . . . To go beyond the asymmetrical ritual of exhibition, the idea that ‘something’ needs a certain degree of agency, even at a minimum a certain indifference of being exhibited, not addressed, but addressing. That ‘something’ is always there but not always available to appearances. That ‘something’ is an otherness, an alterity, an alien that takes up different types of appearance. Being so, it also transforms the visitor into a raw witness.
‘I am not into ‘topics,’ addressing or giving answers, but into perplexity, enigma, weirdness.’—Pierre Huyghe
RK: In this situation at Kistefos, are the visitors to the work in a sense less visitors than some sort of ghosts in a posthuman future? Do they complete a certain narrative by their presence? Is it supposed to feel posthuman?
PH: I’m not thinking posthuman, but rather beyond what we put under ‘the human.’
RK: Meaning what?
PH: I guess some illusionary boundaries and categories that define human exceptionalism, our thought of ourselves being more than nature and machine, having consciousness, language. But to go back to your question: people moving through that place in Kistefos are witnesses. It is really a question of attention to what is around. People are crossing a forest. It is immersive. Moving through, they remark things or not. These things are discreet mutations, sometimes unremarkable things. They do not need to be interesting. Various things might be more obvious than others, say a mutation of a deer skeleton or of beehives. At the end of the island is a large LED screen that displays the real-time simulation based on a scan of the island. An intelligent camera is moving through perpetual change in the simulated environment, hosting the contingency occurring in the actual. You are in the same place that you are seeing on screen, but within it, you can see things mutating, changing. The simulated forest and its actual counterpart are starting to become undifferentiated. As you cross the forest again to exit the island, the attention gets hypertrophic. The natural milieu became augmented. The reality became the actual and also its possibility.
RK: And the parameters for the mutations, or deformations, are set by you?
PH: It’s a set of rules, or prompts, given to an artificial neural network that would, say, in the case of an existing object, a skeleton for example, optimize an image of the skeleton. The optimized image, an alternate idea of a skeleton, is implemented within the simulation.
RK: And what comes out of that the optimization might, to our sensibility, be monstrous, or unidentifiable?
PH: Possibly, as the optimization is an ideal approximation, an alien looking at the actual environment. All the sensory information captured by the sensors goes into the simulation in real time and modifies its shape. There’s always tension between the neural network trying to generate new shapes and the actual environment giving real-time data that is modifying what is generated. This tension can eventually be monstrous, some manifestation of it, physically landing on the actual before disappearing.
RK: And during the course of the exhibition there will be more of these outputted physical mutations, these objects, emerging from the simulation to be made in three dimensions and placed on the actual land to change it?
PH: It’s interesting that you say ‘exhibition.’ I’m interested in making things appear, but not in exhibiting them. That’s why I was talking about something that is unremarkable and indifferent to the public presence. And, in answer to your question, things should exceed the simulation, be outputted and manifested on the island for as long as the project goes. The project is supposed to be ‘permanent’. But of course, you can imagine that there are practicalities and economics involved for a project like this within an art institution.
RK: In a very simplistic sense, this system that you have set up operates, on the physical side, something like a sculpture maker?
PH: I’m not sure about the word sculpture. Maybe a generator of possible things.
RK: And what the machine chooses, to a degree, is not set by you? It’s looking around by itself, making a branch, or a brick, or machine part or a mushroom or whatever?
PH: In an ideal configuration of the work, it operates as you just said. It would have a certain kind of agency, a sympoetic relation, an existence, seen or not. Neural networks (two models, one known as a diffusion model and the other a GAN, or generative adversarial network) generate shapes from existing things on site (including the living), then that mutation is randomly chosen and made into a physical thing, placed at the same place where the thing was in the simulation.
RK: And in an ideal situation, the physical mutations that emerge from the simulation and take physical presence on the land would then themselves be the subject of reinterpretation by the machine?
PH: Yes, as sensors would capture data of their presence.
RK: And, again in an ideal situation, the simulation could then produce a mutation of a mutation, and then that could be outputted to go out onto the land?
PH: Absolutely. If it identifies a mutation on site, it might then mutate the mutation.
RK: And if this went on for five-hundred or a thousand years and those mutations continued to be outputted and placed, at a certain point the physical reality of that island would …
PH: Be completely modified, transformed, contaminated by its possibility. That’s how it is on paper at least, the philosophical and poetic ideal of the project. And then of course comes the reality of how long it can go on and how it can operate.
RK: My last question is about your use of artificial intelligence, machine learning, which has figured into other recent work. How did that interest develop?
PH: When deep learning emerged, and with it the unpredictability in that learning, I was interested in the ways in which inorganic life naturally integrates the work along with organic material, which is also unpredictable. Deep learning is another lifeform, another character. It is not that I do not see the distinction between the two, but I tried to open the moment of undifferentiation between these different types of intelligence. I’m interested in working with something that could feel or perceive or has a certain sentience, a milieu crossed by different contingencies, wherever they are emerging from.
RK: As we all become more aware of how machine learning and artificial intelligence will radically change the way the world works (or maybe won’t work), some of the information that we know about it now can be, frankly terrifying. Is your use of it intended to make us reflect on that, about what A.I. will mean for human life in the coming years? Or is it more just a tool, like any other tool you use to make work?
PH: I would lean toward the second. I am not into ‘topics,’ addressing or giving answers, but into perplexity, enigma, weirdness. One might find here a possibility to think about uncertainty. Somehow, simulations are ways to ride contingency. A parallel milieu to the actual. Different times existing in the parallel milieu, different possibilities of itself nesting within it at the same time.
Pierre Huyghe’s site-specific work ‘Variants’ is now on view in the sculpture park of Kistefos, Jevnaker, Norway.