From Self-Confidence to Statistical Accuracy: Scaling Human-in-the-Loop Evaluation with Direct and LLM-Assisted Auditing

arXiv:2610.35698 · cs.CL · Submitted 2026-10-26 · Read on arXiv