Autodidactic Feedback Loop

A self-correcting cycle where an agent evaluates its own performance against expert-labeled benchmarks and updates its reasoning strategy. It is the engine of recursive self-learning.