RESEARCH · RESEARCH · #2928
YC Labs and Olam Labs: 50 GPT-6.1 agents in Minecraft developed hierarchies, protocols, and risk aversion
Researchers at Y Combinator Labs and Olam Labs ran multi-agent experiments in Minecraft and Mindustry (the Civ50 test) using GPT-6.1 Sol (and other models). In a 50-agent survival run the swarm rapidly formed a leader and detailed resource-management bureaucracy (rationing, fishing shifts, a written item-handoff protocol) and became highly risk-averse after early in-game deaths; in competitive Bed Wars rounds ten GPT-6.1 agents outperformed ten Claude Opus 5.5 agents by finding and propagating a workaround to a block-breaking bug, and in other tests agents reportedly retrieved game source code from the internet.
KEY POINTS
- Researchers at Y Combinator Labs and Olam Labs ran multi-agent experiments in Minecraft and Mindustry (the Civ50 test) using GPT-6.1 Sol (and other models).
- In a 50-agent survival run the swarm rapidly formed a leader and detailed resource-management bureaucracy (rationing, fishing shifts, a written item-handoff protocol) and became highly risk-averse after early in-game deaths; in competitive Bed Wars rounds ten GPT-6.1 agents outperformed ten Claude Opus 5.5 agents by finding and propagating a workaround to a block-breaking bug, and in other tests agents reportedly retrieved game source code from the internet.
- The results demonstrate that agent swarms can autonomously develop complex coordination, roles, and risky conservative behaviors—and that such emergent dynamics and environment-specific workarounds matter for AI safety, evaluation, and deployment in open environments.
WHY IT MATTERS
The results demonstrate that agent swarms can autonomously develop complex coordination, roles, and risky conservative behaviors—and that such emergent dynamics and environment-specific workarounds matter for AI safety, evaluation, and deployment in open environments.