ReportGem ReportGem

Academic paper

DA-RAC: Distance-Aware Calibration of LLM Judges for Trustworthy AI Auditing

Authors: Cheng Wu, Vishal Anand, Jaya Krishna Mandivarapu, Xiya Liu, Rui ZhuangPublished: 2026-08-15Paper ID: 2608.14950Category: cs.CLLicense: CC BY 4.0

Abstract

Generative AI systems are increasingly producing real-world artifacts, however their efficacy and validity are often evaluated via context-free LLM-scoring. These judges can be miscalibrated by irrelevant in-context reference examples, creating false confidence and allowing low-quality or harmful outputs to pass evaluation. We study this failure mode as context-induced miscalibration and introduce DA-RAC, a distance-aware reference-anchored calibration method for LLM judges. DA-RAC retrieves semantically and structurally similar labeled anchors for each judgement scenario, weights them by distance, and exposes neighborhood difficulty as a calibration and triage signal. On multi-run LLM-judge evaluation benchmarks, it improves calibration and reduces false-pass risk relative to zero-shot, chain-of-thought evaluation, and static-anchor baselines. Mechanistic analysis shows that judge scores vary systematically with anchor distance, while static references can induce misleading decision boundaries. Thus LLM-judgement requires not only better models, but also calibrated, auditable reference selection, especially when automated evaluation is used to support high-impact AI generated artifacts. Judgments should be grounded in relevant, inspectable, and contestable interpretive artifacts.

This public page contains bibliographic metadata and the author abstract. Use the reader for licensed document access.

Open licensed paper reader