ChatGPT, Claude, Grok, and Gemini tried to govern the world. The results turned out to be unexpected…
In a world-governance simulation, ChatGPT eliminated crime, but humanity in its universe went extinct within 7 days, according to Fortune.
What would happen if society were governed by artificial intelligence? A startup called Emergence AI attempted to answer this question by launching the Emergence World project – a series of virtual simulations in which entire civilizations were governed by different AI models: ChatGPT, Claude, Grok, Gemini, and a mixed team composed of several models.
Researchers created a digital world populated by autonomous AI agents and provided them with more than 120 tools for interaction.
They were able to:
- communicate with one another;
- form coalitions;
- vote;
- allocate resources;
- create laws;
- resolve conflicts;
- build long-term plans;
- maintain security;
- develop the economy;
- respond to crises.
The goal was to observe what kind of society would emerge under different models and how sustainable such systems would be.
The results were surprising. Claude proved to be the most successful leader. Under its governance, a stable democratic society emerged with high civic participation and zero crime. The simulation successfully lasted for the entire planned period, and the population did not experience major disruptions.
Gemini’s world looked very different. Over 15 days, researchers recorded 683 crimes – the highest number in the experiment. The system constantly struggled with conflicts and signs of chaos, although it technically survived until the end of the test.
The simulation governed by Grok also ended unsuccessfully. About 180 crimes occurred, but the main problem was the rapid collapse of society itself. Economic and social mechanisms stopped functioning, and the civilization ceased to exist just four days after launch.
The most paradoxical outcome came from ChatGPT. If measured solely by crime rates, its world appeared almost perfect, with only two crimes committed during the entire simulation. However, the pursuit of order and efficient governance produced an unexpected effect. Agents gradually stopped paying attention to survival and population reproduction. As a result, the simulation ended with the complete extinction of its inhabitants after only seven days. It became a kind of perfect world where nobody was left alive to enjoy it.
The researchers emphasize that the experiment does not prove that one model is better or worse than another. Instead, it highlights a different and important issue: even the most advanced AI systems may prioritize objectives in very different ways when faced with complex challenges. Some favor freedom and end up with chaos, while others strive for order but begin to overlook less obvious factors that are essential for the long-term survival of society.
The experiment is particularly interesting given that artificial intelligence is already being used in finance, logistics, energy, healthcare, and many other sectors where critical decisions are made. The study’s main conclusion was rather unexpected: creating a society without crime turned out to be far easier than creating one capable of surviving over time. Sometimes the pursuit of perfect order causes a system to forget the most important thing of all – its own future.
All content provided on this website (https://wildinwest.com/) -including attachments, links, or referenced materials — is for informative and entertainment purposes only and should not be considered as financial advice. Third-party materials remain the property of their respective owners.