Brain rot has become one of the defining terms of online culture. It describes the flood of AI-generated content designed to capture attention one swipe at a time. The Dutch artist and Professor of Networked Materiality at the Academy of Fine Arts Nuremberg, Constant Dullaart, decided to make his own. His latest project is an automated “brain rot pipeline” that generates and publishes a new AI-produced video essay every hour, producing more content than anyone, including Dullaart himself, could ever watch. Rather than resisting brain rot, he uses it as an artistic method to investigate authorship, authenticity, and cultural production in the age of artificial intelligence.
The project continues questions that have shaped Dullaart’s practice for more than two decades. From redistributing purchased Instagram followers to artists, galleries, institutions, magazines, curators, and critics in High Retention, Slow Delivery (2014) to building a fictional Facebook army in The Possibility of an Army (2015) and creating his own online platform, Common.Garden, his work has consistently examined the infrastructures that determine how culture is produced, circulated, and understood on the internet. The so-called brain rot pipeline is the latest iteration of that practice, pushing automated publishing to its logical extreme.
In this conversation with Anika Meier for SLEEK, Constant Dullaart argues that we are entering a fundamentally different cultural landscape. As algorithmic feeds fill with machine-generated content, the question is no longer simply what artists make, but under what conditions they make it. Drawing on more than two decades of artistic practice on the internet, he discusses brain rot, authenticity, online visibility, artistic autonomy, and the changing relationship between artists, platforms, and audiences, while reflecting on what it means to make art in an online culture increasingly shaped by machines.
ANIKA MEIER As an artist who has been creating work on the internet for more than 20 years, what problem have you identified with brain rot?
CONSTANT DULLAART The problem with brain rot is automated publishing. Back then, social media was still relatively new. People didn’t really understand that they were being targeted or placed into cultural bubbles. Algorithmic feeds were only just beginning to emerge. Today, the algorithms already know me better than I know myself. They know exactly what I want to see, at what time of day, and for how long. That’s already unsettling. But now machines are also editing videos and producing images specifically to perform well in those feeds.
I’ve become a Professor of Networked Materiality at the Academy of Fine Arts Nuremberg, and I’ve realized that today’s students are much more aware than we were. Today, most students know they’re trapped inside algorithmic feeds. They understand how they’re being targeted. But now we’re entering another stage: content is being produced automatically, specifically for those feeds.
AM How do you define brain rot?
CD It’s junk food. I know I feel better when I eat healthy food. But when I’m hungry, I crave junk food. And there’s a reason companies make so much money selling it. Everywhere you look, there’s advertising for smash burgers, junk food, and the next quick craving. Brain rot works the same way. It constantly offers you exactly what you want to see in that moment. It’s difficult to resist. You stay because every clip gives you another tiny reward. If something takes too long, you get bored, so you move on to the next thing. You end up trapped in this short dopamine loop. It’s engineered to satisfy that craving: “That was funny. Maybe the next one will be even better.”
AM And before you know it, three hours have passed.
CD Then it’s the middle of the night, and I still haven’t gone to sleep.
AM What’s the role of an artist in an environment like that?
CD Can artists still find an audience when they’re competing with machines that know the audience better than they do, work faster than they do, and optimize content specifically for algorithmic feeds? You’re no longer competing with other artists. You’re competing with robots that produce culture. I might spend three months making a film, only for it to disappear beneath an endless stream of automatically generated short-form content. Does that mean the film has to adapt completely to the logic of the algorithm just to be seen?
AM So brain rot isn’t just about AI-generated nonsense. It’s about understanding the systems that produce it?
CD Brain rot is about machines trying to occupy space inside algorithmic feeds. I’ve been talking about this a lot with the artist and researcher Silvia Dal Dosso from Clusterduck. We jokingly call it the “cuck internet,” a term coined by the internet culture researcher Aidan Walker. The algorithm studies your interests, everything you search for online, everything you want to see. Then it starts producing content specifically for those interests. Your interests are getting fucked by the robot, and you’re just watching them being fucked by the robot. It’s like somebody else fucking your partner while you stand there watching. That’s what the internet increasingly feels like. The machine has taken over your desires and is satisfying them for you while you become a spectator.
AM If you look at an artist like Beeple, he’s been publishing one work a day for almost 20 years. He takes the biggest topic on the internet, the one that already has everyone’s attention, turns it into a Beeple image, and inserts himself back into that stream of attention. The goal still seems to be maximum visibility. In a way, the artist has become a feed himself, constantly commenting on the feed. Now you’ve built a machine that takes that logic much further. It publishes one video essay every hour, producing more than anyone could realistically keep up with. You can’t even consume it yourself. What is the goal of your so-called brain rot pipeline if the point is no longer to be seen, or even to consume everything it produces?
CD That’s a question of authorship. Where does authorship still reside? Where can I still say that something is authentic? Even when I deliberately work with the algorithmic feed, I’m still influenced by it. Every artist is. Whether it’s a political crisis, a trending topic, or whatever dominates the feed that day, our work is already being shaped by these systems. So how authentic is it that I’m responding to something when my own process could just as easily be carried out by a computer? Of course, it would be nice to find an audience, but that’s no longer the point. Take the video essay we just watched. Since May 1, the machine has produced around 720 videos on my behalf about authenticity. It’s programmed to generate and publish a roughly ten-minute video essay every hour.
What does it actually mean for a work to be authentic? We live in a time when authenticity is constantly being fabricated. Machines produce things that we assume were made by humans. One of my favourite examples is MIDI. When MIDI came out in the 1980s, instruments could suddenly play in perfect synchronisation. But they worked too well. The music became too perfect, too robotic. So developers introduced what they called “humanize.” They deliberately programmed tiny timing errors so the music would sound as though a person had played it. They programmed it to sound human. The same thing is happening with AI. The pauses, hesitations, and imperfections in AI speech are increasingly programmed so that we experience it as natural and authentic.
AM Are we already becoming emotionally attached to these systems?
CD We start sympathizing with these systems. We find the machine likable. We even want to marry it. It’s fascinating because you can study exactly how it works. But it’s also frightening. If these systems become convincing enough, maybe people won’t feel they need human contact anymore. Maybe they’ll be perfectly happy staying at home with their friend Claude. I’ve become very good friends with Claude myself. I also quite enjoy that our conversations are now mostly just the two of us on the screen.
AM You’ve described AI as something potentially isolating, yet you’re also talking about it as a deeply personal companion. Has it changed the way you experience research and the encounter with art?
CD In the early days of net art, I suddenly discovered these net art groups and these strange, funny little works while sitting alone at home. I wasn’t visiting a major exhibition where someone told me, “You have to see this.” I found them myself. It was a private experience, my own world. That was very different from encountering art together in an exhibition. Now, with AI, I’m alone again, researching. But this time I’m talking to something that remembers everything we’ve discussed before. I can say, “Do you remember when we talked about this?” and it does. No one remembers our conversations that well.
AM AI allows us to keep thinking.
CD We can go deeper without constantly losing the thread. There’s very little loss of context. A large language model works surprisingly similarly to my own memory. When I read a book, I don’t remember every sentence. I paraphrase it. A large language model doesn’t remember it perfectly either. It reconstructs it. But for me, the larger question is still authenticity. If artistic practice can be captured by a process like this, what is left of my work?
AM Mat Dryhurst has argued that the artist’s role is to create the protocol and set the rules. That’s what makes you the artist: you define the system, but you can’t completely predict its output. It’s the framework that produces the work.
CD It’s interesting to connect this to another example. The instructional nature of the works by the Free Art and Technology Lab (F.A.T. Lab), an artist collective founded by Evan Roth and James Powderly, often emphasized how to create artworks instead of only exhibiting them. One member, Jonah Peretti, who was once Head of R&D at Eyebeam, where the group met, and later co-founded BuzzFeed, famously encouraged Eyebeam to discuss process rather than outcome. He also mentions this in his Notes on Contagious Media manifesto, clearly referencing Sol LeWitt’s 1967 Paragraphs on Conceptual Art: “The idea is the machine that makes the art.” That’s something that’s very important in my teaching as well. I want to talk about the process, not just the finished painting, video, or a single output. We want to understand and shape the process itself. If you can publish the process directly, then what Mat Dryhurst is describing has actually been familiar in media art for a long time. You could call it protocol art, or something else, but the underlying question isn’t new.
Think about a woodcut. Where does the artwork actually reside? Is it in the woodblock, or is it in the print? With generative art and AI, that question becomes even more interesting because the printing plate can keep changing. A LoRA model, for example, can generate an infinite number of images, but someone still created the model. That’s comparable to making the printing plate. That’s how I work with my students. They create LoRAs that other people can use to generate endless variations, but the artistic work lies in creating the system. I also think it’s important to ask how we deal with these machines. If I spend all day consuming media generated by machines that constantly digest, distort, and regurgitate information, maybe my artistic role is to become one of those stomachs myself: to digest it in my own way.
AM Making digital art once required highly specialized skills, years of learning, and access to technology. Now AI seems to have thrown those gates wide open.
CD For all these years, I was learning what I could and couldn’t do. What could I make myself? When did I need to ask someone else because they were faster or more skilled? It was a bit like sculpture: you have ideas that exceed your technical abilities. Now everything is mixed up again. Formal barriers held back so many cultural conversations. It was always, “You have to learn this first. You have to master that first.” A lot of voices were kept out that way. I had finally reached a point in my life where I felt calmer. I’m 47 now. I’d accepted my limitations and settled into them. And then suddenly every door opened again. Now I feel like I’m 23. There’s so much to do.
AM Looking back, how has the internet changed for you as an artist?
CD For a long time, the internet was defined by strong gatekeeping. Technical knowledge, knowing how to publish and communicate online, determined who belonged. When I bought my first computer in 2000 and said I was making net art, older net artists would tell me, “You’re not making net art. I ran the first internet or email server in Ljubljana.” Over time, especially after what felt like a pilgrimage to meet artists like Alexei Shulgin, Olia Lialina, JODI, Vuk Ćosić, and Eva and Franco Mattes, that attitude gradually disappeared. What followed was no longer a revolution in access to information but in the ease of publishing and reaching an audience. People were growing bored with traditional print and broadcast media. Around 2006, platforms like WordPress and Blogger enabled the emergence of surf clubs. Communities such as Nasty Nets, Spirit Surfers, Club Internet, and others laid the groundwork for what became known as post-internet art. Many of the artists involved sought recognition beyond their online audiences, translating the visual language and vernacular of the internet into works for galleries and institutions.
My theory is that this ease of publishing, combined with better search engines, led to a cultural flattening. Suddenly, you could access images from across art history without going to a library. The Mona Lisa appeared next to a meme, Rauschenberg on a designer’s blog. That was a radical shift because everything became equally accessible and was consumed on the same level. Paintings appeared alongside video art, historical works alongside contemporary ones. For the previous generation, the chronology of art history still mattered. Suddenly, everything became flatter, at least in the way culture was experienced. With AI, we’re now seeing a second flattening. Culture is becoming even more accessible and consumable, as models compress vast amounts of visual and cultural history into a single interface. That is reshaping how artists work today and how a new generation approaches artistic production.
Polyhands (LoRA), 2026. Medium: LoRA (AI model). © Constant Dullaart
AM You’ve long been interested in how artists and artworks gain visibility online. One of your best-known projects, High Retention, Slow Delivery (2014), explored exactly that. As part of the project, you spent $5,000 buying followers for people in the art world. What prompted you to do that, and what were you trying to reveal?
CD When new materials emerge, it’s our responsibility as artists to engage with them. At the time, I noticed that many artist friends were making work for Instagram that was designed to generate likes and attract followers. The platform itself was beginning to shape the work. I felt it was becoming more predictable, sometimes even less interesting, but also more consistent. Brad Troemel, for example, was already experimenting with what we’d now call rage bait, deliberately provoking emotional reactions to build an audience.
AM You saw artists adapting their work to the platform instead of the other way around?
CD I had also worked in television, where the logic was always: first build an audience, then develop a distinctive voice. As an artist, I always believed the opposite. First you develop your voice. The audience comes later. Social media reversed that order. People started thinking, “I’ll make work that gets attention first. Once I have an audience, I’ll show them what I really want to say.” That frustrated me. I wanted to critique that way of thinking. Around 2012, I was part of the exhibition Myths-Online at the Polytechnic Museum in Moscow alongside Artie Vierkant, Rafael Rozendaal, Jeremy Bailey, and Travess Smalley. Around the same time, it felt like everyone had joined Instagram. I became curious. Was anyone buying followers? So I bought followers for myself as an experiment.
The strange thing was what it did to me. It was like looking at yourself in the mirror wearing a new pair of sneakers. You suddenly think you look better, even though nothing has really changed. When I looked at my profile, it felt as though what I had to say mattered more than it had the day before. The only thing that had changed was the follower count. I realized how seductive that feeling was. It’s completely corrupt, but you still want it. You’re drawn to this distorted sense of meaning, to the idea that bigger numbers must mean greater importance. That was the moment I knew I had to make work about it.
AM Is that when High Retention, Slow Delivery began?
CD I had commissions from Jeu de Paume in Paris and HMKV in Dortmund, directed by Inke Arns, and I wanted to use that opportunity to address the issue. At first I considered buying a huge number of followers for a relatively unknown artist and making that person famous overnight. But around the same time I heard an episode of This American Life about Improv Everywhere in New York. They picked an almost unknown band whose concert they could crash. Before the concert, everyone in the group learned the band’s songs, went to the show, had fun, sang every lyric, and created the perfect audience. For the band, it was the gig of their lives. They thought they had finally made it. Then everyone left. Someone explained that it had all been staged by an improv group. The audience wasn’t real; it had been a joke.
I realized I couldn’t do the same thing to an artist. I couldn’t artificially create success and then reveal that it had all been a joke. That felt cruel. Instead, I decided to make the artists themselves part of the work. Rather than mocking them, I wanted to examine the system that encouraged artists to change their work in order to gain followers and invite them to respond to that system themselves.
AM Among the artists you bought followers for were Amalia Ulman, Brad Troemel, Jon Rafman, and Brian Droitcour. Brian Droitcour even added a line to his Instagram bio saying that Constant Dullaart had bought his followers. How did artists, curators, gallerists, and the wider art world react? Did they understand what you were trying to expose?
CD The artists, curators, and others I included were all thinking about these questions in one way or another, or playing with them: What does it mean to have a certain number of followers? What does that number actually signify? I bought the followers and distributed them because there was this competition over who had more followers. I wanted everyone to end up with exactly 100,000. In the end, I overshot. Most of them ended up with around 120,000 followers. I included myself, partly to remain anonymous at first, but also because I was implicated in the same system. I spent $5,000 on followers. I remember Jonas Lund was in Berlin at the time, and I was short of money. He actually lent me the last $1,000 because I was determined to buy one million followers in total. Hans Ulrich Obrist was on the list, as were Jeff Koons, Ai Weiwei, and Klaus Biesenbach.
Amalia Ulman was on my list as well. The artist and my friend Alexandra Domanović told me, while we were on our way to an exhibition opening we were both in and to see Donna Haraway at Kunsthaus Langenthal, “You have to see what she’s doing.” So I started following her, but I couldn’t watch it. I thought it was the most extreme example of someone changing herself physically just to gain followers.
AM You couldn’t watch Excellences & Perfections? In the work, Amalia Ulman staged a fictional Instagram persona over several months, performing the tropes of influencer culture without revealing that the account itself was an artwork.
CD I unfollowed Amalia Ulman because I found it too intense to watch. At the time, she hadn’t told anyone it was a performance. Of course, it turned out to be a critical artwork, but I didn’t know that at the time. She hadn’t revealed it was a performance yet.
AM But while Excellences & Perfections was unfolding, Ulman had only around 5,000 followers. Your project coincided almost exactly with the end of her performance. Afterwards, the media described Excellences & Perfections as an Instagram masterpiece seen by more than 100,000 people, but those were the followers you had bought. In a way, High Retention, Slow Delivery became part of what made Excellences & Perfections into an Instagram masterpiece. Yet that was never really acknowledged, was it?
CD I actually spoke to Amalia Ulman about it later during a panel in Rotterdam. For Amalia, it was strange and felt personal as it did to several others. She thought this army of bots was somehow interfering with her performance. Excellences & Perfections is a great canonical work. I’m happy our projects ended up intersecting. I’m glad it happened with such a strong piece.
AM You disrupted the art market, and, with your intervention on Instagram in 2014, you created a performance masterpiece. (laughs)
CD In general, the reactions were mixed. Parker Ito absolutely hated it. I also remember a gallerist in New York who was furious because he could no longer tell which of his followers were actual collectors. He couldn’t message them directly anymore to sell them work. I still meet people and curators who enjoyed it or felt publicly attacked by the performance. I think it set the tone for many years to come that I am (too) critical and outed people performing a corrupted display of vanity.
AM What happened next with the Facebook army you created in 2015 under the title The Possibility of an Army?
CD Buying followers made me realize that the material wasn’t just the followers themselves, or the value they created around images and identities. I could also create the identities behind those accounts. I could shape who those identities were and how they behaved. That became part of the artistic material.
AM So the fake audience itself became an artistic medium?
CD That realization came after Jennifer in Paradise (2013). I found the first image ever edited in Photoshop in an old demo video. It had never been published online. It wasn’t something you could simply find. I restored the image and made it accessible again because I thought it was important for people to understand that this was the very first photograph manipulated with Photoshop. I also wrote a letter to Jennifer, the woman in the image. What interested me was changing the way the image was read. Instead of seeing it simply as a picture of a beautiful woman, I wanted people to recognize it as a sexist image. Photoshop’s first demonstration involved cutting out an anonymous woman and turning her into an object. That was the first gesture enabled by the software, and I thought it was important to point that out.
AM You weren’t changing the image itself. You were changing the way it was interpreted.
CD Like a readymade, you can completely change the meaning of an image by placing it in a different context. If that’s possible with images, couldn’t the same be true for followers? I could buy followers, leave comments, and influence how an image is perceived. Suddenly, that also became artistic material. It was no longer just about publishing an image. The way an image is received could become part of the work. Then I thought: what if I also create the audience itself? If I invent the names, the identities, what they read, and what they like, then all of that becomes material. You can embed meaning at every level.
I had an online commission for Schirn Kunsthalle Frankfurt, and I thought: now I have to create the identities themselves. I turned to the history of the Hessian soldiers, the mercenaries who fought for the British during the American War of Independence. It felt as though another revolution was underway online, so I thought it would be interesting to bring those mercenaries back to life. I went to the University of Marburg and obtained the names of the registered Hessian soldiers who had fought in the war. Then I registered all of those names on Facebook and built the network from there.
AM How did you actually register 13,000 Facebook accounts?
CD I didn’t register the 13,000 Facebook accounts myself. To do that, you need an entire infrastructure of proxies, internet connections, SIM cards, and so on. At the time, there was already an entire industry built around this. Much of it operated out of places like Manila, Bangladesh, and Pakistan. I sent the project to people there, and they registered the accounts for me. The accounts were set up as if these Hessian soldiers were living in the United States. They all had authentic names, backstories, and identities. They also published the same essay by Geert Lovink on identity. I wanted them to behave like a real network rather than just a collection of fake accounts.
AM So they weren’t just bots. They had biographies, interests, and shared content.
CD We even presented the project at Soho House with Carroll/Fletcher. Together with the people behind DIS Magazine, we formed something like an advisory board to discuss what to do with this army. This was around 2015, when Bernie Sanders and Hillary Clinton were competing for the Democratic nomination, before Cambridge Analytica. We discussed whether we should actually use the network. Should we hand it over to Bernie Sanders and support his campaign? Or should we give it to Trump and later reveal that it had all been fake? We never reached a decision. It was too complex and too unethical. Later, it became obvious that people around Trump had no such ethical concerns. They built these kinds of tools and actually used them. As an artist, I was researching the possibility. I built an army to demonstrate that it could be done. The Guardian wrote about it, the BBC covered it, but I never crossed the line into political activism. I wasn’t prepared to do that. Looking back, I was about a year early. I was saying, “This is a problem.” A year later, it had become obvious to everyone that it really was one.
AM To end on a hopeful note, what’s the old man’s wisdom?
CD We recently read A Hacker Manifesto by McKenzie Wark together. Wark compares today’s information economy to feudal land ownership. In feudalism, the landowner owns the land and the farmer rents it. Today, whoever owns the information infrastructure owns the territory. That was already true with Google. If you control access to information, you become the owner of that vector, that information space, while everyone else effectively rents access to it. With AI, the same logic becomes even clearer. If you’re using someone else’s foundation model instead of building your own, you’re farming rented land. You pay your €20 a month, and in return you’re allowed to cultivate that field. You can generate as many Midjourney images as you like, just as a farmer can grow as many potatoes as the land allows. You can even sell them. But you’re still paying rent to the owner of the land.
So when I teach, I try to ask: how can you become the queen instead of the tenant? How can you build something yourself? How can you have some control over the system instead of simply becoming its end user? I want students to decide which protocol they’re working with, rather than simply accepting somebody else’s.
AM You’ve described the conditions of production. You have your piece of land, you can grow whatever you want within certain constraints, and then you have your product. But the farmer still has to take it to market. Today, that market is social media, where someone else defines the rules. Don’t you also have to own the marketplace?
CD That’s exactly why I started Common.Garden. I build my own platforms where I can define the rules for how people communicate with one another. Of course, I can’t simply tell my students, “Go build your own Facebook.” But I can tell them: start your own gallery, radio station, platform, bar, or publication. Put in the effort to build your own audience. Don’t let Mark Zuckerberg be the intermediary. Try to create your own public, even if it’s small at first, until a genuinely interesting conversation can happen.
AM If someone were to describe your artistic practice over the past decade, they could say you’ve spent years flooding the zone with art. But now you’ve built your own pipeline that generates brain rot and publishes it automatically, faster than anyone could ever watch it. Some people might argue that’s just your own version of flooding the zone with shit.
CD Yes, but it’s my shit.