Dataland, billed as the world's first Museum of AI Art, opened June 20, 2026, in downtown Los Angeles, and it works differently from any museum you've visited. The art inside doesn't sit still. It evolves in real time, driven by two streams of data: live ecological information from rainforests around the world and biometric readings taken directly from visitors wearing wristbands. The museum's own site describes the concept plainly: "data becomes pigment, and art continuously evolves in real time, shaped by human interaction, machine intelligence, and the vast but invisible information that shapes our world."
Dataland: A New Paradigm for AI Art
The museum is the creation of digital artist Refik Anadol, whom Spectrum News California notes was named by Time magazine as one of the most influential people in AI. It occupies a building designed around five multi-sensory galleries, and the engineering firm Arup, which worked on the project, says the technologies are integrated into the walls, ceilings, and floors so the infrastructure effectively disappears into the architecture. The result, in Arup's framing, is a building that becomes "an active participant in the experience" — a living museum that "dreams and remembers."
That framing is more than a slogan. The museum's inaugural show, Machine Dreams: Rainforest, is described by its organizers as an attempt to translate the intelligence of the natural world into interrelated sensory experiences. Because most people will never stand inside a rainforest or hear its tens of millions of bird songs, Anadol uses machine intelligence as a means of access, preservation, and amplification.
Dataland's Living Artwork: The Feedback Loop Explained
Here is the step-by-step model of how the system works, assembled from the museum's own description, Arup's engineering account, and Spectrum News California's reporting. The loop runs continuously: inputs feed the AI, the AI generates art, the art changes the visitors, and the visitors' changing bodies feed the loop again.
| Loop stage | What happens | What the evidence specifies |
|---|---|---|
| 1. Visitor input | Data.Link wristbands read each visitor's body | Heartbeat and emotion monitored through changes in the skin — "goosebumps, to be more precise," in Anadol's words |
| 2. Environmental input | Live ecological data streams in from rainforests worldwide | Weather reports from the rainforest, plus datasets gathered across 16 rainforest environments and nature repositories like the Encyclopedia of Life |
| 3. AI processing | The Large Nature Model synthesizes both streams | An open-access multimodal AI model created by Refik Anadol Studio, based solely on nature data and hosted on Google Cloud |
| 4. Artistic output | The model generates evolving art across the galleries | Real-time visuals across 3,500 square feet of LEDs and 1.5 billion pixels, soundscapes, and scent augmentation |
| 5. Visitor response | Visitors react to what they see, hear, and smell | Their physiological responses flow back through the wristbands, restarting the loop |
One honest caveat: the sources describe the loop as a whole, but none of them details the exact algorithmic mapping — how a specific biometric reading, such as a spike in goosebumps, translates into a specific change in color or sound. The feedback relationship is documented; the precise input-to-output correlation is not.
