Consensus across multiple AI models on a particular output or judgment, used as a signal of correctness.