Frontier Research
Ambitious projects that push the Institute's research from theory into working systems.
Postulates 3, 4, 5
Rebuilding the system that changed AI, from scratch.
Reproducing AlphaGo from the ground up: MCTS, deep neural network evaluation, and self-play RL. The goal is to understand how an AI system learns to make decisions under uncertainty.
Postulates 1, 2, 3
What emerges when game design starts from first principles?
Deriving a new strategy game from axioms. Start from desired mathematical properties like complexity, depth, and emergent strategy, then construct rules that produce them.
Postulates 2, 3, 4
A programming language for formal, axiomatic thinking.
A research programming language designed around formal structure, explicit relations, and composable knowledge, where programs are knowledge representations.
Postulates 2, 3, 6
A laboratory for market mechanism research.
A simulation environment for testing prediction market mechanisms, agent strategies, and information structures under controlled conditions. Build the markets, then watch them work or fail.