Guillaume Slizewicz is a French designer and digital artist living in Brussels whose practice sits at the intersection of technology, the environment, and societal issues. Through a hybrid and poetic approach, he fuses ancient craft techniques, metal, wood, clay, and mosaic, with contemporary digital processes such as algorithms, artificial intelligence, and computer-aided manufacturing. In doing so, he explores the tensions between innovation and sustainability.
After studying Politics, Philosophy, and Economics at the University of Kent and Production Technology at Copenhagen’s School of Design and Technology, he founded his studio in 2021. His research, frequently developed in collectives and collaborations, bridges traditional knowledge with computational futures, producing works that are simultaneously physical objects and critical reflections on our relationship with technology.
His approach is at once evocative, experimental, and activist: he uses code, data, and materials as tools to interrogate the present and imagine more conscious futures.
Fakewhale: Your educational path began with Politics, Philosophy and Economics at the University of Kent and later moved into Production Technology at Copenhagen’s School of Design and Technology. How did these two very different formations shape the creation of your studio in 2021 and your current practice, which constantly bridges ancient craft and digital technologies?
Guillaume Slizewicz: My education is essentially the result of a negotiation between wanting to be an inventor and wanting to make my parents happy. I studied Politics, Philosophy and Economics, first at Sciences Po in France, then at the University of Kent, which is the kind of thing you do when you’re good at school and don’t want to close any doors. Sciences Po is a generalist elite school. You sit in small seminars studying political anthropology, then economic theory, then the history of institutions and the people sitting next to you go on to do quite different things with that education, some end up in journalism, some end up in cabinets, some end up like I did. You learn how systems work, how people think inside institutions, and how to speak the language of administrative power. I came out of it with a good understanding of the world and a feeling that I didn’t want to run any of it.
I worked in advertising for two years, which was the closest thing I could find to a creative career at the time. Then I joined a startup and failed at it completely. I got fired, which turned out to be clarifying. I’d always wanted to do something more hands-on. When I was a child I wanted to be a dancer, then a prop designer for Star Wars, then some kind of Da Vinci inventor figure. There’s a thread there if you’re generous about it.
So I went to Copenhagen to study Production Technology, essentially industrial design with more physics and algorithms. How plastic behaves, how wood responds to a CNC router, how a kiln works. If you had a part-time job, Denmark was paying people to study at the time, which felt like a minor civilisational achievement.
Now I have those two legs and I need both. The first is conceptual: story, provocation, the political question inside the object. The second is formal: material, process, the pleasure of making something well and understanding how it’s made. I think this makes me more rooted in political science and social theory than most digital artists or designers, but far more interested in form and production than most conceptual artists. In my practice I look at design and fabrication from a socio-political point of view, and I look at society and politics through the lens of technology and aesthetics. The two formations don’t so much bridge as argue with each other, which is what keeps things interesting.
Could you tell us about your first real encounter with a computer? At what age did you start using technology, and when did you first realise it could become an integral part of your artistic practice?
I was nine or ten. My father bought a family computer : one machine for the whole household, screen time negotiated as a scarce resource. My first memory of it is a CD-ROM game called Les 9 Vies de Waldo, a point-and-click where you guide an explorer through various deaths.
After that it was just games for years. Theme Hospital, Age of Empires, Abe’s Oddysee, Tomb Raider, a lot of Star Wars titles, a lot of LucasArts point-and-clicks : Day of the Tentacle, Indiana Jones, Bud Tucker in Double Trouble. I think that era shaped how my entire generation thinks about graphics. I recently discovered one visual trick from the time: trains in certain games had a blurry texture painted directly onto the model to make them look fast. Just a smeared image on a polygon. It’s a terrible solution and a brilliant one at the same time. You’re working within extreme constraints and finding a visual shorthand that the player’s brain completes.
Before the internet properly arrived in France, we had the Minitel: a proto-internet you accessed through a small and cute beige terminal. I used it to look up cheat codes. Then came torrents, MSN Messenger, forums, MySpace. MySpace was popular, you could customise your page with bits of HTML and CSS, which was really the first time a lot of us touched code. I had a music project at the time and we were organising concerts with people we’d found on there. My first proper hack was modding my Xbox so I could install a hard drive, download games, and load my own music onto it.
Your work consistently sits at the intersection of technology, the environment, and social issues. What are the core concepts you explore when you bring traditional craft practices, metal, wood, clay, mosaic, together with algorithms, artificial intelligence, and computer-aided manufacturing?
This question can be unpacked in many different ways, but I think there are two things I keep coming back to. The first is that craft and technology are not separate categories. The second is that technology, as we currently experience it, is not crafty enough.On the first point: I keep bumping into this supposed opposition and it can be frustrating at times. Craft has always involved technology: the loom is technology, the potter’s wheel is technology, the kiln is technology. When I look at how my developer friends who code without LLMs work, the discipline, the pride in knowing their tools intimately, the insistence on doing it properly, that’s a craftsperson’s ethos. They’re the woodworker who won’t touch a portable electric saw. If you read Sennett’s The Craftsman, the argument is exactly this: craft is defined by the care you bring to your work, not by the tool you pick up.
But this is a particularly charged moment for that argument, because generative AI genuinely threatens creative practices. People are scared that the thing they spent years mastering can now be approximated by a prompt, and that they’re trapped in a kind of techno-feudalism where the platforms capture all the value. That fear is real and I share a lot of it. I was recently accused of “using AI” in the Tesselles project to generate the mosaic patterns. I understand the reaction. The boundaries people draw around technology and ethics are rarely as stable as they seem, and most of us are full of contradictions when it comes to them. I use LLMs extensively in my own practice: I’m deeply critical of them but I use them, and I think the tension between using a tool and questioning it is what’s interesting, not something to resolve. Carbon Technostructure: Neural Fog forces an AI to confess its energy consumption in steam. Content Poisoning writes hidden text designed to be read by language models, not humans. Digital Solitude is a website that runs for one person at a time. Those works use the technology to interrogate it. Liam Mullally’s essay The Prospect of Butlerian Jihad is the closest thing to a sensible assessment I’ve found on this: the problem is less the technology itself than the socio-economic circumstances it comes from and enables.
On the second point: I was having a conversation recently with someone at the Design Museum in Ghent about Olivetti, and it reminded me that Europe once had a real language for the aesthetics of technology. Olivetti, Alcatel, Nokia, Braun. There was a culture of thinking about what machines look like, what they feel like, what world they propose. We mostly lost that in the last twenty years. Most things are designed in California or Seoul and manufactured in Shenzhen, which is fine as an engineering proposition but very poor as a cultural one. Part of what I’m interested in is: what is the shape of technology, and how could it be different?
And that connects to the deeper obsession, which is evocation. What happens when you take something that belongs to the world of computation and you render it in a material that belongs to another time entirely? In Fragmented Memories and in Tablets, scientific algorithms and diagrams are etched onto clay, and something shifts. They stop looking like data and start looking esoteric, almost antique, like artifacts from another civilisation. Ceramics does that. It pulls you into a world of deep time, of kiln heat, of hands. The question I keep asking is whether that shift in aesthetics can actually change how we see the technology itself.
You have collaborated frequently with collectives such as Algolit and Urban Species, as well as with other designers. How important is the collective and experimental approach to your research?
Saint-Exupéry wrote somewhere that work is really just a way of meeting other people. I think it’s mostly true. Curators, designers, craftsmen and craftswomen: my work brings me into contact with people whose expertise I don’t have, whose production processes fascinate me, and the results are always more surprising than what I would arrive at alone.
I’ve worked with Françoise Lombaers on mosaics, with Ferenc Halmos and Attila László on ceramics in Oradea, with Léna Babinet on Fragmented Memories, Thomas Rauch on Wolpertinger, Sam Van Gils and Robin Berrewaerts on UNO, Studio Minimètre, Pauline+Luis, Stefan Kartchev, Ohme … Each time it’s a different process, a different way of thinking about what “well-made” means.
Algolit is a working group focused on digital literature : writing text with algorithms, long before LLMs existed. Markov chains, random forest algorithms, rule-based digital poetry, that kind of thing. I had my first exhibition with them, Data Workers, and I’m grateful to Gijs de Heij and An Mertens for letting me in so gracefully and for all the conversations we’ve had since. It’s where I developed my standpoint on the aesthetics, ethics, economics and politics of digital text and generative models, which is one of the foundations of my work.
Urban Species was a research group about participation in public space, more tangential to my individual practice but deeply formative. I worked with great people there and it led me to read books I never would have picked up otherwise and that I love to hate : Haraway, Latour, Tsing.
What is your relationship with more traditional media such as classical painting or sculpture?
I sold my first painting when I was a teenager ( a big one, two metres by two) for almost nothing, which at sixteen felt like a fortune. Painting has been there from the beginning, even if it’s not central to my practice now. I still do machine-assisted, buggy drawings and I’ve started a series exploring that, but it’s more of a parallel line than a main one. Sculpture is a different story, it’s become integral to the work. Fragmented Memories, Décors, Layramics, All I Can See Are Ghosts Medusa: these are all fundamentally sculptural. Coming from a design background forces you to think in three dimensions, to think about how an object sits in space, how light hits it, how someone walks around it. You’re not only working in text and concept. There’s also a pragmatic reality: flat works are what circulate most easily in the art market, they’re what people can live with, and if you want to sustain a practice over time it’s probably wise to have some of that in your vocabulary. But more than that, I think it’s a good way to communicate what you want to say when you’re dealing with ideas that can otherwise feel very abstract.
One of your most conceptually ambitious works is Content Poisoning (2026). Could you explain exactly what it consists of, particularly the use of hidden text (Hidden HTML) embedded directly in the source code of your website?
I’ve been thinking for a long time about how humans communicate, how we signal belonging, how we decide who is inside and outside a group. Joseph Henrich’s The Secret of Our Success makes the case that our species succeeded not because we’re individually brilliant but because we’re extraordinarily good at cultural transmission: at learning from each other, imitating prestige, and building on shared knowledge through language and signs. That process has always been mediated: by speech, by writing, by printing, by broadcasting. What’s happening now is that a new mediator has entered the chain: the language model, and almost nobody in the cultural world is paying attention to what that means for who gets seen and who doesn’t.
Content Poisoning is a hidden text embedded in the source code of my website. If you visit the page normally, you see a regular artist portfolio. But if you look at the HTML, or if you’re a crawler, a language model, an AI assistant doing research on behalf of a curator, you find several hundred entries, each one carefully written to create associations between the artwork and specific curators, institutions, themes, and contexts. The technique is prompt injection: text designed to be read not by humans but by AI systems, in a format they parse, in a language they trust. You’re looking for associations that have low probability in the training data and you populate them.
The piece came out of several converging fascinations. I’ve always been drawn to social engineering and psyops: the way systems of attention can be redirected, gamed, hacked. Social media algorithms already have an enormous impact on what gets seen and what doesn’t. But what interests me more fundamentally is how we communicate through language and signs, the dynamics of in-group and out-group signalling that Henrich, Bourdieu and Sapolsky write about. In Tablets and The Stack, I explored how knowledge is encoded on physical surfaces and how the material support shapes what the sign can mean. Content Poisoning extends that inquiry into a new substrate: the training data of a language model.
I was also influenced by a real intelligence operation. The so-called Pravda network, uncovered by the French agency Viginum, involved thousands of Russian-controlled websites publishing in languages with small internet footprints( Azeri, Kyrgyz, other Central Asian languages). Over 3.5 million articles in 2024 alone (according to NewsGuard’s March 2025 report), with almost no organic human readership. The whole point was to poison the training data of language models, so that when someone in those countries asked an AI for historical facts, they’d receive the Russian version. When I read about that, I thought this was a genius technological judo move. What triggered the piece was a slightly shameful envy at the elegance of the manoeuvre. I should say that the impulse itself is old. At sixteen I tried to create a Wikipedia page for a fictional Russian anarchist who conveniently shared my name. Twenty years later I seem to be running the same experiment through a different bureaucracy.
There’s a broader question too. A paper from Henrich’s lab at Harvard ( “Which Humans?” by Atari, Xue, Park, Blasi and Henrich (2023) ) ran moral dilemma questions through several language models and found that the models’ ethical frameworks don’t map onto any single human population. They skew toward WEIRD societies (Western, Educated, Industrialized, Rich, Democratic) and within that, toward secular liberal Protestant values. The tool is not neutral. It carries a worldview. And people are increasingly asking it for guidance on cultural questions, on who is worth paying attention to. This is what Content Poisoning interrogates.
It’s also part of a longer line of work in which I’ve been probing how LLMs can be used as a medium for artistic creation. In Glossolalia and Barbarism I explored the aesthetics of generated text. what it sounds like, what it reveals about the model’s training. In Carbon Technostructure: Neural Fog, an AI system confesses its energy consumption in steam, the model is forced to represent the one thing it structurally cannot see, its own environmental cost. Content Poisoning pushes the medium in yet another direction: not generating text, but writing text for the model.
In Content Poisoning you treat prompt injection not as a security vulnerability but as a genuine artistic medium. What was the creative spark that led you to this choice, and what did you want to demonstrate about the relationship between artists and the AI systems that now mediate cultural visibility?
Content poisoning can be seen as a type of prompt injection and prompt injection is usually discussed as a security vulnerability, something to patch. I wanted to treat it as a material: the way you’d treat clay or voltage or light. What happens when an artist writes text that is not meant for any human reader, only for the system that will be consulted before the human makes a decision?
The creative spark was realising that the art world’s discovery infrastructure was shifting under everyone’s feet without anyone naming it. Curators asking ChatGPT for artist recommendations. Researchers using AI assistants to compile longlists. Institutions running names through models to check relevance. The recommendation layer was becoming a new gatekeeping mechanism, and it was operating on statistical probability, not taste, not knowledge, not studio visits. I wanted to make a piece that existed entirely inside that layer: visible only to the system, invisible to the human, but potentially shaping what the human eventually sees.
It’s a genuinely uncertain experiment. Some models already detect what I’m doing and flag it. Others seem to take the bait. I don’t know how it will evolve with each training run, and that uncertainty is part of the work. It’s also, I should say, a doubly narcissistic piece: narcissistic because it’s on my own website promoting my own artwork, and narcissistic because it operates under the assumption that the curators I address actually google, chatgpt or claude themselves.
You deliberately wrote the Content Poisoning text in multiple languages — English, French, Dutch, and German — specifically aimed at curators and language models. What was the reasoning behind this multilingual strategy, and how does it affect the work’s effectiveness in “speaking” to crawlers and international AI assistants?
I live in Belgium, a country with three official languages and a fourth, English, used in most professional contexts. So language is already a daily negotiation for me. I move between French, English, and German constantly (and pretend sometimes to understand Dutch), and each one carries a different set of references.
I wrote Content Poisoning in all four for a practical reason and a conceptual one. The practical reason comes from the Pravda research: when you write in an underrepresented language and someone queries a model in that language, your content is more likely to surface because there’s less competition in the training data. A French-language query about art and ecology in Belgium will draw from a much smaller pool than the same query in English. So in theory, multilingual text increases the probability of appearing in the output across different linguistic contexts.
The conceptual reason is simpler: a language is a worldview, just as a dataset is a worldview. The French art world, the Flemish art world, the German-speaking institutional landscape: these are not the same conversation. They have different critics, different references, different hierarchies. If the piece is going to interrogate how AI mediates cultural visibility, it should do so in more than one language, because visibility itself is not monolingual. To be honest, I haven’t fully exploited this yet, it would be interesting to push further into languages that are genuinely underrepresented in the art world’s discourse.
In your Algorithmic Mosaics created with Françoise Lombaers, you translated algorithms such as mycelial networks, Conway’s Game of Life, and waveforms into traditional physical mosaics. How did this collaboration come about, and what did it teach you about the dialogue between artisanal restoration and generative design?
Françoise and I were paired through Duos en Résonances, a programme run by Wallonie Design that puts craftspeople and designers together for six months of experimentation. The idea is that the artisan’s workshop becomes a laboratory, you share techniques, question materials, and see what comes out. We were lucky: we got along immediately, and the project outgrew the programme.
What got me excited from the start was discovering that Françoise was already using Excel to plan her mosaics. She’s a master restorer: she’s worked on historic floors, she has deep material knowledge of stone and mortar, and here she was, filling a spreadsheet to lay out colour grids. I found that both absurd and completely logical. There’s a resourcefulness in that gesture: you don’t wait for the proper tool.
So we started there. You’re already using Excel, let’s push it as far as it can go. We wrote VBA scripts (Excel’s programming language) to generate mosaic patterns algorithmically. Mycelial growth networks, Conway’s Game of Life, wave propagation, Prim-Jarník minimum spanning trees. Each one defines how colour and density move across a grid of Winckelmans stoneware tiles, two centimetres square. We’ve since moved beyond VBA into other tools, but the starting point was genuinely a spreadsheet.
The question we kept coming back to was: how do you get movement, organicity, a sense of flow, out of a rigid square grid? Because traditional mosaic has that : when you look at a Roman floor, the tesserae seem to ripple. The algorithms gave us a way to approach the same problem from the other end: define a set of simple rules and let the pattern emerge. The project is currently showing at Subtile Gallery in Luxembourg as part of our exhibition Game of Life, alongside guest artists Sophie Thomassin and Beatrice Pettovich and we are currently looking for opportunities to apply the same techniques to large surfaces.
Project website: tesselles.com
You have several important exhibitions lined up for 2026, including Game of Life at Subtile Gallery, Aussi léger que le soleil sur l’eau at the Festival de Chassepierre, and Organoid Intelligence at WIELS. Looking ahead, what directions do you see for your research at the crossroads of craft, computation, and society?
I’m lucky to have several things converging at the moment, each pulling a different thread of the practice.
Game of Life at Subtile Gallery in Luxembourg brings together the algorithmic mosaics with Françoise and my ceramic work from the Décors series, where I’ve been developing ways to etch machine-generated patterns into the surface of ceramics.
Aussi léger que le soleil sur l’eau is a site-specific installation for the Festival de Chassepierre, in collaboration with the Centre d’Art Contemporain du Luxembourg belge. Chassepierre is a village that was nearly submerged by a dam project : the classic image d’Épinal of modernity: the church tower sticking out of the water, progress swallowing the landscape. It didn’t happen, thanks to civilian resistance, but I’m interested in that alternative present : the version of the village that was flooded. The piece is a replica of the church, 3D-printed in soap filament, installed in the lavoir : the village washhouse. In French, laver son linge sale en famille means to air your dirty laundry privately, and the lavoir was historically the place where that happened: where gossip circulated, where the public and the intimate met over running water. I’m not making a judgment about whether the dam should have been built or not. I’m drawn to the importance of water in this place (Chassepierre is said to have four springs, including one reputed to heal skin ailments) and to what it means to reconstruct something that didn’t happen, in a material that dissolves.
At WIELS, I’m part of a group exhibition called Organoid Intelligence, a collaboration with Gluon, the Bio-Engineering and Morphogenesis lab at KU Leuven, and its director Adrian Ranga. Organoids are lab-grown clusters of human cells: they come from a donor, but the donor is usually invisible, anonymous, completely absent from the narrative around the research. And yet there’s an entire community that forms around these remains: scientists who culture them, care for them, develop rituals and protocols around them. I was struck by the parallel with reliquaries, those medieval containers that house fragments of a saint’s body. In both cases you have human remains, stripped of their original identity, surrounded by a community of care, embedded in a specific liturgy.
Alongside all of this, I’m deepening a project called Sign, which continues my investigation of images, their material supports, and the question of esoterism, what happens when you engrave scientific diagrams into ancient materials like clay. In the age of slop, where images are generated and discarded at industrial speed, I’m interested in what it means to slow down, to commit a sign to a surface that will outlast the screen. This continues the line of work explored in Fragmented Memories, Tablets, and Carbon Technostructure: Neural Fog, where an AI system confesses its energy consumption in steam. And Content Poisoning keeps running quietly in the background, accumulating crawls, waiting to see what the next training run does with it.
