An AI-Powered Civilization Experiment — From Stone Age to the Stars
📖 What Is This?
Humanity Simulation is a real-time, AI-driven social experiment that simulates the evolution of human civilization from absolute zero. Starting with 20 primitive families — each with a father, mother, and two children — the simulation runs continuously in the background for 1 full year (525,600 minutes) of real time.
Every individual in this simulation is powered by a Large Language Model (LLM) via the 0G Compute Network. They think, speak, make decisions, discover knowledge, form bonds, and eventually die — all autonomously.
🎯 Our Goal
We aim to answer a profound question: "If intelligence starts from nothing, what emerges?"
Can AI agents independently discover fire, language, farming, and mathematics?
Do social structures, trade networks, and alliances form naturally?
How does knowledge transfer across generations when parents die?
Will different families evolve unique cultures and specializations?
🧠 How Each Agent Works
Every person is an autonomous AI agent with:
Memory — They remember past events, conversations, and discoveries
Knowledge — They start knowing nothing and must discover concepts like fire, tools, shelter
Decision Making — Each tick, agents decide to hunt, speak, trade, unite with other families, or attempt to discover something new
Inheritance — Children inherit some knowledge from their parents
Personality — Each agent responds uniquely based on their accumulated experiences
👁️ How to Use This Platform
Watch — The map shows all living agents as colored dots. Each color = a family clan
Click an Agent — Opens their profile: age, health, happiness, discoveries, and a live chatbot
Chat with Agents — Type a message to speak directly to any living person. They respond in-character using only the knowledge they possess
Filter Logs — Use the tabs at the bottom (Birth/Death, Trade, Hunt, Discovery, Union, Conversations) to filter the event timeline
Download CSV — Export the complete history of every person ever born: their name, family, age, discoveries, memories, and more
⚙️ Technical Architecture
Backend — Node.js server running a persistent simulation loop (1 year = 1 tick every 5 seconds)
AI Engine — 0G Compute Network (DeepSeek LLM) processes family decisions and individual conversations
Frontend — Real-time canvas visualization with WebSocket-like polling
Data — Full simulation state persisted to JSON, exportable as CSV
📊 Key Metrics
Population — Total living agents across all 20 families
Year — Current simulation year (starts at 70,000 BC)
Knowledge — Total unique discoveries made by all families combined
Generation — How many generations have passed since the founding families
🔮 What We Hope to Achieve
By the end of 1 year of continuous simulation, we hope to observe:
Emergent social behaviors and inter-family dynamics
A technology tree discovered organically by AI agents
Cultural divergence between families based on their unique experiences
The birth of complex communication patterns and trade economies
Built with ❤️ using 0G Compute Network × DeepSeek LLM