sim-ai.xyz

Experimental Simulation Lab

EXP-001

Environment Genesis

Generating physical environments from parameter seeds. Each simulation begins with a random initialization vector that blooms into a unique world -- terrain, atmosphere, physics constants, all emergent from a single seed value.

STATUS: RUNNING
META-001

Seed: 0x4A9EFF

Iterations: 2,847,391

Convergence: 94.2%

EXP-002

Agent Emergence

Deploying learning agents into generated environments and observing emergent behaviors. No pre-programmed rules -- agents develop strategies through pure environmental interaction. Every run produces different societies.

STATUS: COMPLETE
META-002

Agents: 10,000

Generations: 450

Novel behaviors: 23

EXP-003

Reality Drift

Testing the boundary between simulation and training data. How far can a simulated reality diverge from its training distribution before agents can detect they are inside a simulation? Preliminary results: further than expected.

STATUS: RUNNING
META-003

Drift: 0.847 sigma

Detection: 12.3%

Confidence: High