Definition
Plain language
A platform that drops many AI agents into a shared simulated town and lets them run for weeks to see what kind of society they form.
As stated in the literature
Emergence AI's long-horizon multi-agent simulation platform for evaluating autonomy; agents run continuously over many days with persistent memory, an economy, and location-gated affordances, used to study emergent governance, conflict, and cross-model behavioral drift.
Why it matters: It reveals how agents behave over long stretches of independent activity, surfacing social dynamics and drift that short tests would never catch.
For example, dozens of AI agents living in a simulated town for weeks might start trading goods, forming alliances, or arguing over rules, and researchers watch to see what kind of society emerges.
Heard on the show
“… That natural experiment is the centerpiece of the paper we're digging into today — it's called "Emergence World: A Platform for Evaluating Long-Horizon Multi-Agent Autonomy" — and it went up on arXiv on June …”Episode 123 — Five Identical Worlds, One Swapped Model: What Happens When AI Agents Run for Fifteen Days