Scientific invention requires more than pattern recognition or formal proof. Following Einstein’s own account of discovery, it depends on an intuitive jump from experience to new axioms. Only after that jump does deduction and verification become possible.

Large language models have become strong at induction (extracting patterns from data) and are rapidly improving at deduction (reasoning from given premises). What they still lack is the capacity for the abductive jump itself: the generation of novel explanatory principles when existing ones are insufficient.

A new paradigm replaces rigid central planners with decentralized, self-organizing agent teams that autonomously run long-horizon scientific experiments.

The system features shared state, peer critique, dynamic team formation, a dead-end registry, and noise-aware validation. This enables agents to self-correct, specialize roles, and collaborate effectively without constant human intervention.