Critique Loop

A mechanism that enables an AI agent to perform structured self-review of its own outputs. The agent first generates a solution, then the model critiques that solution and iteratively revises it. Conceptually, this parallels the actor-critic architecture found in reinforcement learning.