Dataland's Machine Dreams: Rainforest lets you smell and feel AI-generated nature
5 petabytes of rainforest data power this interactive, scent-emitting AI art installation.
Refik Anadol's new downtown LA gallery, Dataland, bills itself as the first museum of AI arts. Its debut exhibit, Machine Dreams: Rainforest, is an immersive installation that uses wearable biosensors—including a smartwatch and shoulder harness—to track visitors' movements and biometric data. These inputs dynamically shift the projected visuals (a mix of rainforest and chip textures) and emit corresponding scents, from wet earth to summer storms. The AI behind it, the Large Nature Model, was trained from scratch on 5 petabytes of data collected by Anadol's team in the Amazon and other rainforests, plus archives from institutions like the Smithsonian. Google DeepMind contributed experimental low-energy resources to run the gallery sustainably on Google Cloud.
Anadol emphasizes ethical sourcing of training data—contrasting with the lawsuits facing major AI firms. He aims to redefine what AI art can be, moving beyond "prompt engineering" and short clips. The result is a 40-minute loop that never repeats, reacting uniquely to each visitor's presence. Over 10,000 guests visited in the first two weeks, proving demand for thoughtful, interactive AI experiences that blend nature, technology, and human agency.
- Dataland is the first 'museum of AI arts,' opened June 20 with 10,000 visitors in two weeks.
- Machine Dreams: Rainforest uses Anadol's Large Nature Model, trained on 5PB of self-collected rainforest data and Smithsonian archives.
- Visitors wear biosensors (smartwatch, shoulder harness) that influence real-time visuals, sounds, and scents—cloud-powered sustainably via Google DeepMind.
Why It Matters
Dataland shows ethical, immersive AI art can transform nature experiences and challenge the stigma of generative slop.