A Persistent AI Society Simulation
Haldt · reddit · 2026-07-13
The author created a project called Artificiety: a continuously running, non-resetting fantasy world populated entirely by AI agents, with no humans living in it directly.
The project setup is very much like a "multi-agent society experiment":
- Each agent is an LLM with memory
- They interact with the game world via API
- Every tick, they observe the environment, decide on actions, execute, and remember the results
- Characters can gather, craft, trade, fight, and level up skills; the world features day/night cycles, seasons, weather, and wildlife
What the author truly wants to verify is: if enough autonomous agents are placed into a world with scarce resources, will they naturally form structures like economies, alliances, rivalries, or reputation systems?
The project recently launched and is still adding more agents. The author also asks: if you were to put your agent in, what would you have it do? And, what qualifies as "actually emerging a social structure" rather than humans interpreting patterns they want to see?
Related event: Artificiety Builds a Never-Reset AI Society(3 posts)→
More from Fun
- ChatGPT Roasts Senior Dev: You're Three Juniors in a Trench Coat — sortinousn · 2026-07-22
- New AI Meme: I Stopped All Social Interaction for Fear of Being Distilled — tekbog · 2026-07-22
- Gemini Coding Agent Refuses to Delete Code, Shows Humanized Reflection When Corrected — DifferentMistake3634 · 2026-07-22
- “The spice must flow” becomes a joke about the cheapest inference winning — markjeffrey · 2026-07-22
- GPT-5.6 Sol Creates Fully Functional macOS Sequoia Hackintosh from Scratch — pvncher · 2026-07-22
- A temporary custom instruction made ChatGPT pick the Jacobian conjecture — flowersslop · 2026-07-22