← Retour au blog
tech 20 May 2026

Growing Neural Cellular Automata: The Evolution of Morphogenesis

Growing neural cellular automata are revolutionizing our understanding of morphogenesis through differentiable models capable of growth, persistence, and pattern regeneration. Let's explore how these models are shaping the next generation of self-organizing systems.

Article inspired by the original source
Growing Neural Cellular Automata ↗ distill.pub

Introduction

In the captivating world of artificial intelligence and self-organizing systems, Growing Neural Cellular Automata (GNCA) represent a major breakthrough. Inspired by morphogenesis—the process by which organisms develop their shape—these models offer valuable insights into how complex structures can emerge from simple rules.

Understanding Neural Cellular Automata

Cellular automata are systems composed of cells that evolve on a discrete grid over time, according to defined rules. GNCA add a layer of complexity by integrating neural networks to dictate these growth and regeneration rules. Similar to natural morphogenesis, these models can generate complex, stable patterns capable of regenerating after disruptions.

Growth, Persistence, and Regeneration

Growing Models

Growing models are trained to generate patterns, but their long-term persistence is not guaranteed. Despite this, some patterns stabilize or even partially regenerate. This illustrates the potential robustness of simple rules.

Persistent Models

Persistent models, unlike growing models, are designed to maintain patterns over extended periods. This stability is crucial for applications requiring memory or constant structure. They also exhibit regenerative capabilities, even without explicit instructions.

Regenerating Models

Lastly, regenerating models are specifically trained to withstand damage. Their ability to regenerate patterns, especially in central areas, represents a major technological leap, paving the way for systems capable of self-repair.

Applications and Implications

GNCA offer revolutionary perspectives across various fields. In synthetic biology, they could serve as a foundation for designing organisms capable of self-repair or adapting to their environment. In computing, these models can inspire new data processing architectures, where systems are capable of learning and adapting in real-time.

Challenges and Future Perspectives

Despite their promise, GNCA pose challenges. Their increased complexity requires significant computational resources and a robust theoretical framework to fully understand their behavior. However, with ongoing advances in AI and computing power, these challenges are surmountable.

Conclusion

Growing Neural Cellular Automata embody the future of self-organizing systems. By drawing inspiration from natural biological processes, they pave the way for unprecedented innovations across various technological and scientific fields. Let's discuss your project in 15 minutes.

automates cellulaires morphogenèse intelligence artificielle systèmes auto-organisés régénération
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call