What Kexin Wang (Brown University / Rhode Island School of Design) has built looks at first like a strange printer. A petri dish sits beneath a laser head. The dish holds a layer of nutrient agar; fungal spores are scattered across it. The laser moves back and forth, repeatedly tracing an outline onto the surface—not to etch, not to illuminate, but to repel. Fungi are photophobic organisms: short-wavelength lasers pose a continuous threat. The organism refuses to grow into any area the laser has touched. Forty-eight hours later, the traced interior remains empty, while dense mycelial networks press up to its boundary on the outside. The shape you see is the exact impression of where life refused to go.

Behind this sits a complete computational apparatus. Wang's team first trained an image segmentation model (E-ViT) and a temporal convolutional network (TCN) on time-lapse footage of actual fungal colonies, learning the species-specific spread dynamics from the images themselves. These learned patterns are then compressed into a neural-network-driven cellular automaton—each simulated cell runs its own small network, dynamically deciding whether to expand based on the states of its neighbors. The simulation engine, written in Rust, sustains thirty frames per second with a million active cells. When the automaton predicts that a region is approaching nutrient exhaustion and the fungus will soon stop spreading, the system automatically cuts the laser there; it maintains the boundary only where active spread is still occurring. The forbidden zone therefore contracts dynamically over time, and the fungus always chases the space it is allowed to inhabit. The system has been validated across four species—Rhizopus oligosporus, Phanerochaete velutina, Physarum polycephalum, and Aspergillus niger—each producing a distinctly different texture of spread, a different quality of mark.

In the SAE framework, what is happening here can be described precisely. The already-construct of art-making is: the tool as executor. Brush, chisel, inkjet head, pen plotter, algorithmic renderer—the shared logic is that intention is traced onto a passive substrate. The substrate has no will; the tool carries intention; their contact leaves a mark. Even generative art follows this logic: the algorithm executes rules, the screen or paper accepts the output, the substrate remains passive. This logic has been in place since cave walls, through laser engravers, and has never been fundamentally disturbed. Wang's chisel is to replace substrate-passivity with organism-agency. The laser does not draw—it forbids. The fungus does not submit—it avoids. The form—the thing we call the work—is precisely the record of a living thing making autonomous decisions inside a geometric prohibition. The maker is neither the artist nor the machine but life acting under a boundary constraint.

This exposes a remainder that has existed throughout art history but has never been named: the use of a living organism's autonomous avoidance as a forming technique. This is not Bio-Art in the tradition of Eduardo Kac or Oron Catts, where life is cultivated, modified, or displayed as subject or medium—where life is what gets chiseled. In Funguy, life is the chisel: it makes choices, it avoids, it negotiates with the boundary the laser imposes, and the final form is the crystallized residue of that negotiation. Nor is this generative art, where the generativity depends on algorithmic execution—code runs, pixels or vectors emerge, and the human is only the rule-designer. Funguy's central actor—the fungus—executes no human-written rule. It runs an avoidance program written by hundreds of millions of years of evolution, a program no human code can see, control, or accurately predict (only approximate). Nor is it laser engraving, whose logic is that where the laser touches, a mark is left. Here the inverse holds: where the laser touches, there is nothing. The artwork exists only where the laser never reached.

Wang's paper, published in SIGGRAPH Asia 2024 Art Papers, carries the title "Exploring Fungal Morphology Simulation and Dynamic Light Containment from a Graphics Generation Perspective"—a purely technical description that does not touch the aesthetic core of the paradigm. Hackaday's July 2026 coverage tags the project "fungal art" and "laser control," both accurate, but their conjunction cannot describe what is structurally new here. The naming gap itself is evidence that the remainder exists: critical language has not yet caught up with what is happening. Right now, funguylab.com has just launched. The catalog of available species is still expanding. Each new species brings a new spread grammar, a new quality of mark, a new negotiation style. The logic has not yet closed. It is still growing.

See it now. When "constrained bio-form" or "avoidance aesthetics" becomes critical vocabulary—when it occupies a paragraph in a media art history—that moment will already be construct: a once-strange structure that has fully digested itself. But right now, standing beside the petri dish and watching it grow—watching a living thing grow around the place it cannot go, using its own presence to draw the shape of its own absence—this remains an unsettling, not-yet-named beauty.

Han Qin (秦汉) · @hqinjarsy · SAE Judgment & Aesthetics