Adaptive◉∿◉Gridding at Rondo Sztuki

What is it like to be online in 2026? Fatigue. Stress. Less desire for sex. Excitement. Reverie. Collective euphorias. It’s hot, sticky, pulsating. The terrain trembles slightly, and the grid trembles with it. The demons of the internet, capricious creatures, tempt us from its every nook. Some steal our attention, others lead us by the hand through digital gardens, occasionally rewarding us with serendipity. The landscape is chaotic: data-centre arsons, charming DIY cyberdecks, generated images of war, monumental world models, one-person startups powered by an army of sometimes rogue AI agents.

In places the terrain slides away: models collapse under the pressure of synthetic data, and images begin to document events that never happened. What seemed stable changes before our eyes. The grid constantly adapts to these shifts: it densifies where greater precision is needed, directing more computing power to exactly those areas. An old accelerationist saying goes: the only way out is through. Along the way, however, we look for other paths and temporary shelters. Or, on the contrary, we want to speed up and make the journey even wilder. To see where it leads us.

The exhibition’s title, Adaptive Gridding, refers to a computational method in which the density of a mesh dynamically adjusts to the complexity of the area under study. Within the exhibition, this mechanism becomes a model for responding to the instability of contemporary digital reality. The works by Martyna Marciniak, Sam Lavigne, Sara Bezovšek and Bogna Konior, together with a joint installation by Natalia Kaftan, Maru Drobczyk and Ziemowit Jarecki, propose distinct yet overlapping strategies: mapping, sabotage, full immersion, straying from the main path into the internet’s dark forest, and intercepting and modulating circulating signals.

In Desert of the Unreal, Martyna Marciniak examines the proliferation of synthetic images. She replaces the question “Is technology to blame?” with another: “What is this image a symptom of?” The starting point is the Great Cascadia Earthquake and Tsunami, a disaster that was supposed to occur in 2001 but never happened. In 2023, however, it came into being retroactively through AI-generated images published on Reddit, producing an uncanny sense of collective false memory. This fictional scenario grows out of a real threat: the last major earthquake in the region occurred in 1700, and the possibility of another has long stirred the collective imagination.

In Slow LLM, Sam Lavigne goes a step further — he slows down the machine moving across the terrain. The work takes the form of a browser extension that slows the generation of responses by language models. The artist draws on workers’ methods of resistance: sabotage, the destruction of tools and the deliberate slowing of work, including the “Italian strike”, asking about their contemporary digital equivalent. Instead of destroying the machine, he takes away its pace. This subtle yet exasperating gesture strikes at the basic promise of commercial models: faster, easier, more productive.

The regime of optimisation that Lavigne disrupts at the level of the machine shifts, in Bezovšek’s work, to the body. In After American Psycho, presented as a single-channel video of a website recording, she conveys the seductive and unsettling experience of being extremely online, reconstructing the film’s plot through images that have taken on a life of their own — combining found memes and GIFs with contemporary looksmaxxing content.

For Konior, images become signals of uncertain source and intent. In a video essay commissioned for New Media Week, philosopher and writer Bogna Konior develops themes from her book The Dark Forest Theory of the Internet, presenting the internet not as an agora but as an ongoing first-contact scenario — an alien invasion unfolding in real time. A joint installation by Natalia Kaftan, Maru Drobczyk and Ziemowit Jarecki gathers signals from all the works in the exhibition and returns them to circulation, turning the building into an information stack in which architecture directs the flow of data.

There is no single correct route through this terrain. We can map its distortions, slow down the machines operating on it, immerse ourselves in its images, hide our position, or intercept and modulate circulating signals. Adaptive gridding means the constant need to correct one’s own position as conditions change. The grid moves with the ground, and each of our movements changes its layout.

What is it like to be online in 2026? It means wanting to stay and to leave at once. To commune with something beautiful and terrifying. Larger than you and me.

Kat Zavada