nom nom (2026) online prototype view 
commission by LAS Art Foundation in collaboration with Google Arts and Culture

nom nom


→LAS Art Foundation
→Google Arts and Culture
→Somerset House
→Oxford
→NeurIPS (Publication)




  • World Building
  • Distributed Interaction
  • Collective authorship and intelligence 

nom nom is an artificial world composed of millions of particles. Each particle perceives only its immediate neighbours and updates itself through a tiny neural network.In the work, no single particle possesses knowledge of the whole. Each changes its state based only on interactions with its immediate neighbours, triggering them to grow, collapse, compete or recover. From these local exchanges, landscape-like structures emerge. 

The project draws on Neural Cellular Automata (NCA), a class of systems inspired by how biological organisms grow from simple cells. Rather than programming specific creatures, landscapes, or behaviours, the artist defines local rules of interaction and allows complexity to emerge from them.

The result is neither a game nor a simulation of an existing ecosystem. It is an experiment in constructing a world whose inhabitants, weather systems, and landscapes arise from the collective behaviour of countless microscopic agents.

  • Online Portal
  • Interactivity
  • Generative Landscape

For its first edition, nom nom will be presented on Google Arts and Culture: https://artsandculture.google.com/preview/experiment/-wGAP_EYEvH37w 


- A 30-day simulation: Witness the world regenerate every 24 hours, evolving through local exchanges of Neural Cellular Automata, autonomous creatures and Gaia - an AI-weather system.

- Gaia (Google Gemini): Integrated as an observer, Gemini dynamically alters weather conditions and analyzes the landscape’s evolution with onscreen text-based provocations

- Digital archive: The simulation concludes with a permanent record, allowing visitors to explore the entire month’s transformation.
    • Artifical Ecology
    • Growth, Transformation and Decay
    • Metabolism


    Artificial Ecology is the framework we use to consider nom nom as a generative environment rather than a system for producing isolated images.

    Across these snapshots, diverse forms emerge from the interactions of millions of computational pixel, competing territories, shifting weather conditions, and accumulated histories within the simulation. 

    Each image records a temporary configuration, showing an ecology continuously producing variation throughout one single simulation run. The resulting diversity is showing the visual exploration and possibilities can be generated through the ongoing relationships between digital agents, environments, and time.



    Touring exhibitions:

    Somerset House - 3 Evenings (Oct 2026)
    NeurIPS - World’s leading AI conference - Creative AI track  (Dec 2026)
    Stephen A. Schwarzman Centre for the Humanities, Oxford - Planetary Health exhibition (April 2027)
      



    Timelapse of nom nom’s simulation

    Our Team
    Xin Liu Studio — artistic direction and delivery
    Google Arts & Culture and LAS Art Foundation — original prototype commissioners
    Somerset House — curatorial, production and audience support
    Oxford Cultural Programme — exhibition, research and public-programme context
    Technical and fabrication collaborators — software, installation