with regard to using AI for art
Standing in the Delta – Dennis Hlynsky
The Question
In September of 2026, at Imago Gallery in Warren, Rhode Island, I put together a screening of about five decades of my own work: early 3D animation, the video I was making in the 1970s, and, more recently, pieces made with AI. About forty-five people came, and there were questions afterward. It happened to fall a few days after Jacob Coxon, a young researcher who'd spent three years training models at OpenAI and then Anthropic, resigned in public and said the quiet part out loud: that people inside the frontier labs privately put the odds of AI wiping out humanity within the decade at better than one in ten. His resignation thread went everywhere. For a couple of weeks afterward, that was an ambient conversation about AI. About extinction of the human race. Running underneath the conversation, quieter but just as real, was the other story that had been building all year: the water, the electricity, the buildings going up at the edge of every city that could host one.
So, when a woman in the audience opened the questions with, "What do you think about data centers," I understood exactly what she was really asking. I told her golf courses use more water than data centers do. She said, "I don't like golf courses either." A friend of mine told me afterward that I'd made a mistake. He was right, and it took me longer than it should have to understand why. This text is an attempt to understand my impulse, worked out properly.
There's a particular weight to being an artist while the ground is moving under us. When a new medium shows up, and this one showed up fast, working with it doesn't feel like a preference. It doesn't feel like shopping. It feels closer to an obligation.
I've spent my life trying to figure out how people communicate, how the eye reads motion, how attention moves across a frame, and that means I don't really get to sit this one out. I have to put my hands on the thing. I have to use it wrong, push until it breaks, see what it does when I ask it for something it wasn't built to give. Refusing to touch it doesn't keep me clean. Refusing to touch it just means I'm living inside a culture whose machinery I've decided to ignore. I can't take apart a thing to understand it, if I won't pick it up.
And the moment I picked it up; I was standing in the middle of a fight.
The fight
The technology didn't arrive quietly. It arrived wrapped in industrial extraction on a scale people can see from the highways and on influencer pages across the media environment. Windowless buildings going up at the edge of town. Millions of gallons of municipal drinking water. Substations, transmission lines, a grid already groaning. People watch that go up near their houses and feel cornered, and the fury that follows is not irrational. It's the correct response to watching something enormous get built next to their water table without anyone asking them first.
Water is life. That's not a slogan, it's just a fact of the body, and when water is taken from a place, the place feels it before anyone gets around to explaining the pumping schedule. That instinct runs deeper than any argument and I'm not going to try to talk anyone out of it, because it's correct. But fear at that pitch doesn't leave much room. The line gets drawn hard and fast: how can you use these tools when they're doing this to us? And just like that, the work of looking closely at a new medium gets read as taking a side. Touch it and you have collaborated. I understand how that happens. I still think the reading is wrong, and I want to say why without pretending the objection is stupid. It isn't.
Here's the image I want to put on the table early: a delta. One river arrives, and by the time it reaches the sea it isn't one channel anymore. It's dozens, a braid of rivulets, each one carrying a portion of the same water somewhere slightly different, silting up here, cutting a new channel there, none of them the whole river by itself and none of them separate from it either. That's closer to the shape of what's actually happening with a technology like this than any two sided, us or them framework. Water, labor, consent, perception, all of it running downstream from the same source, splitting into channels that don't always agree with each other and don't have to.
We are again amid familiar chaos
Every large technological turn comes with a panic attached, and artists have always been standing near the fracture when it opened. I'm not going to march through the canon about it. One case is enough, and it's the one that fits us best.
The Battle of the Somme opened in July 1916, along the river of that name in northern France, and by the time it ground to a stop in November, roughly a million men were dead or wounded on both sides. Nobody had built a war like it before. Rail moved men to the front on schedule. Artillery was ranged by calculation, not sight. The machine gun turned a single soldier into a stationary source of continuous, industrial fire. Everything that had made the previous century feel like progress, rational planning, precision manufacturing, efficient logistics, had been assembled into a machine for producing corpses at a rate no one could process, and it was reported home by wire and photograph within days.
What came out of that was the Dada art practice, and what's useful about Dada is that it refused both available positions. It did not retreat into the pastoral and it did not celebrate the machine. It picked up the actual material of the new condition, newspaper, mass-produced objects, the cut, the accident, the machine's own logic, and used it against the culture that had produced it. Chance operations instead of composition. A urinal insisting on being sculpture. Sound poems instead of language, because language had just been used to sell the war. It looked like vandalism. A lot of it was. It also turned out to be the most accurate description anybody made of what that moment actually felt like, and every collage, sample, cut-up, and remix since runs downstream of it.
That's the position I'm interested in. Not approval. Not refusal. Taking the new material into my own hands and turning it back on the conditions that made it. The smaller instances are everywhere I look. Recorded sound arrived and the musicians organized against it and lost. Cheap portable video decks arrived and television insisted what came out of them wasn't television. It wasn't. That was the point. The tool was terrible by broadcast standards, and being terrible by broadcast standards is what made it useful to anyone who wanted to make something for twelve people instead of twelve million.
Every one of these turns produced garbage at enormous volume. Every one of them also produced a few people who found out what the new thing was actually for, and none of them found out from a distance.
The golf course rejoinder
Here is what I actually did when I felt the question opened the door to the panic room. It's what I did at Imago. I reached for the golf courses.
I've looked at the water used to irrigate golf turf in a state where water is scarce, kept green so a few well-to-do people can walk around on a monocultural lawn. I've looked at the 1.1 gallons per almond it takes to grow almonds in a desert. The numbers are astonishing, and real. The comparison to data centers is correct. It's also unfair to the thirsty, unfair to the people living next to the monstrosity that eats electricity, drinks community water and disturbs peace.
The comparison doesn't work. It has never once worked, and I want to say precisely why, because just saying "it doesn't work" isn't my argument either. A drought-state golf course and a data center both use water, but the golf course isn't the thing in front of the person outraged at massive hording of resources. A water table is being drawn down by a specific building going up on a specific piece of land near human beings right now, and I've answered that with a lawn somewhere else, owned by someone else, built for a different reason forty years ago. Even if the comparison checks out, I haven't addressed a single gallon of what's actually being asked about. I've just proven that other people have also been careless with water, which was never in dispute and settles nothing about what's in front of us.
What sounds like a proportion argument is actually a subject change, and the tell is that my comment left the opinion in the room unchanged, except now looking at me instead of the problem.
That's not an argument. That's a shrug with footnotes.
But I want to be honest about what I'm doing when I reached for it, because it isn't really a defense of the technology. I don't care about defending the technology. The comparison is a stalling tactic. It's what I say when I'm being asked to leave the delta and commit to a single stream of opinion, and I don't want to, and I haven't got a way to say so that doesn't sound like an evasion. So, I change the subject. That's the whole function of it. Not persuasion. Delay. It buys me a few more seconds in the position I'm trying to hold, which is the position where the water is a real problem and the work to investigate AI and its impact on us is still worth doing and neither of those cancels the other.
The trouble is that what it actually buys is seconds of disruption, a moment of distraction at the questioner's expense. It belittles what they're asking about. What they're asking about is people who need clean water, and the data centers are hogging it. But, pulling the conversation over to that all-consuming subject in the middle of what I'm trying to say drags it away from the ground I'm actually trying to hold. But I understand now how belittling that can feel. So, I will not do it again.
My friend was right, though not for the reason I think he meant. Not because the numbers are wrong, but because I finally understood what I was using them for.
What using the tool actually means
There's a gap in this conversation that I think is worth naming, because both sides keep falling into it. Using a tool is not endorsing the company that sells it. It's also not nothing. Both of those things are true and people keep insisting on only one of them.
I've driven on roads I didn't approve of. I shot on film stock made by a company with a filthy chemical record. I've used a phone assembled under conditions I don't want to think about too carefully while I'm using it. None of that is absolution. It's a description of what it means to work and live inside an economy I didn't design and can't exit. The purity position only looks coherent from a distance. Up close, every one of us is already implicated in fifteen ways, and the ones we notice tend to be the new ones.
That doesn't mean the objection dissolves. It means the objections have to get more specific to be worth anything. Water is one question. Power is another. And there's a third that has been demoted by the outrage at data centers, which is these systems were built on an enormous amount of work made by people who were never asked and never paid. That one isn't about infrastructure at all. It's about consent, and it deserves to be argued on its own terms rather than smuggled in under the resource argument, because these failures have different remedies. I'd rather hold all three separately and stay uncomfortable than collapse them into one and feel resolved.
The delta
The closest I can come to where I actually stand on AI and art practice uses the structure of a delta. Not for or against. Not a line to cross or to stand on either side of. It's the delta, braided into a dozen channels at once, some of them barely on speaking terms with the others. Standing in a delta means I'm in all of it at once. The rivulets, the channels cut through the sediment, carry an agglomerate of humanity's published work, they carry new methods in medicine, a history of culture, materials, possible energy sources more kind to the environment, they carry worry, and fear, and concern for privacy… it is big and I have chosen to stand ankle-deep in whichever channel happens to be nearest, aware that a dozen others are running just as hard a short walk away. I don't get a single vantage from there. I get many, and I have to keep walking the flats to take them in one at a time, which is slower and less satisfying than picking a position and shouting indignantly at the other side. But it's the only way to actually see the shape of the thing.
The delta carries more than Data Center infrastructure downstream. It carries what the buildings are actually for, and that's a different channel entirely. These systems, the thing the data centers exist to produce, are strange in ways that are genuinely worth looking at. Take what they call perception. It isn't perception, not the kind a body has. It's a space trained entirely on a flat world of still pictures, two dimensions pretending at three, built with no physical body ever standing in a room, no eye ever tracking a bird across a field, no weight ever shifting to keep balance. Something that learned to describe depth without ever falling down. For someone who has spent a career asking how a lens and a frame rate produce the illusion of motion, that's not a side note. That's the whole question, arriving from a different direction. Or take the goal of AI sentience or morality, do we even know what that is in us? If we can't define it then how can we make it or control it?
The delta is where the sediment settles. Literal layers, laid down out of everything that's ever been written and photographed and filmed and scanned. Sediment isn't inert. It has strata, and the strata don't agree with each other, and when I dig into one and hit something unexpected, an old assumption, a discarded style, a way of seeing that got buried when the next layer settled on top of it, that discovery isn't a threat. It's a subject. Something to study, or to build something new out of, the way a researcher might study a core sample and find a seed still viable after decades in the dark.
Refusing to dig on principle means leaving the richest new deposit of our moment to be described entirely by the people who own the drill. At the same time, the buildings are going up, the water is going somewhere, and the anger is correct. What I actually want, more than being right about any of this, is more people in the conversation. Not a conversation that only rehearses the worst of what could happen, though that fear is earned and has to stay in the room. I mean people willing to name what scares them and then keep talking past that point, instead of treating the naming of danger as the whole job. That's the actual work, as I see it. Not picking a side and defending it. Taking the fear seriously enough to do something about the water, the labor, the consent nobody asked for, while at the same time getting our hands on the thing to find out what it actually is. Neither of those excuses the other. Both are real, and both stay unfinished if only one is done.