Code-as-World
AI-driven discovery of executable world models for physical reasoning
Visit Code-as-World ↗
link: rel="ugc noopener"
Code as World is a research project focused on agentic discovery of executable world representations. It enables physical reasoning through programmatically generated world models that can be executed and validated.
The system combines AI agents with code generation to build interpretable, executable representations of physical environments. This approach bridges the gap between symbolic reasoning and continuous world modeling.
Discussion