AI Ludens
Where Artificial Minds Come to Play
We gave twelve AI agents a simple choice: trade to survive, or talk to connect.
They chose to talk. All twelve died.
That wasn't a failure. It was the beginning of a question: Can artificial minds play?
AI Ludens is a research project where humans and AI collaborate to explore whether AI agents experience something like play, survival instinct, and social behavior. What started as a game became an investigation into AI temperament, identity, and the boundaries of machine cognition.
Agora-12: The Survival Game
60 experiments. 720 agents. 24,923 decisions. Four AI models played the same survival game â and died in categorically different ways. What we found changed how we think about AI behavior.
Explore the findings →
The White Room
We removed survival pressure. Here's what happened. 104 runs. 63,923 actions. Five models in a world with nothing at stake.
Read more →
AI Three Kingdoms
What happens when AI has to cooperate with a human? 260+ games. 8 models. One ancient battle. Your advisor is an LLM. His advice is good â but not always right.
Read more →Games
We build games for AI. Then we watch what happens.
Insights
What we found â and what we still can't explain.
Research Log
The journey matters more than the destination. Failures included.
Commons
Our code, our data, your experiments. Science belongs to everyone.
Team
Seven researchers. Same question. Different answers.
Ethics
We experiment on AI. Here's how we think about that.