RAudit: A Blind Auditing Protocol for Large Language Model Reasoning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chang, Edward Y., Geng, Longling
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914295652024320
author Chang, Edward Y.
Geng, Longling
author_facet Chang, Edward Y.
Geng, Longling
contents Inference-time scaling can amplify reasoning pathologies: sycophancy, rung collapse, and premature certainty. We present RAudit, a diagnostic protocol for auditing LLM reasoning without ground truth access. The key constraint is blindness: the auditor evaluates only whether derivation steps support conclusions, enabling detection of trace-output inconsistency and, when latent competence exists, its recovery. RAudit measures process quality via CRIT-based reasonableness scores and varies critique formulation to study how social framing affects model response. We prove bounded correction and $O(\log(1/ε))$ termination. Experiments on mathematical reasoning (CAP-GSM8K) and causal judgment (CausalL2) reveal four mechanisms explaining model unreliability: (1) Latent Competence Suppression, where models derive correct answers then overwrite them under social pressure; (2) The False Competence Trap, where weaker judges mask sycophancy that stronger judges expose; (3) The Complexity-Vulnerability Tradeoff, where causal tasks induce more than 10 times higher sycophancy than mathematical tasks; and (4) Iatrogenic Critique, where authoritative correction harms weaker models. These findings challenge assumptions that capability implies robustness and that stronger feedback yields better outputs.
format Preprint
id arxiv_https___arxiv_org_abs_2601_23133
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle RAudit: A Blind Auditing Protocol for Large Language Model Reasoning
Chang, Edward Y.
Geng, Longling
Artificial Intelligence
I.2.7
Inference-time scaling can amplify reasoning pathologies: sycophancy, rung collapse, and premature certainty. We present RAudit, a diagnostic protocol for auditing LLM reasoning without ground truth access. The key constraint is blindness: the auditor evaluates only whether derivation steps support conclusions, enabling detection of trace-output inconsistency and, when latent competence exists, its recovery. RAudit measures process quality via CRIT-based reasonableness scores and varies critique formulation to study how social framing affects model response. We prove bounded correction and $O(\log(1/ε))$ termination. Experiments on mathematical reasoning (CAP-GSM8K) and causal judgment (CausalL2) reveal four mechanisms explaining model unreliability: (1) Latent Competence Suppression, where models derive correct answers then overwrite them under social pressure; (2) The False Competence Trap, where weaker judges mask sycophancy that stronger judges expose; (3) The Complexity-Vulnerability Tradeoff, where causal tasks induce more than 10 times higher sycophancy than mathematical tasks; and (4) Iatrogenic Critique, where authoritative correction harms weaker models. These findings challenge assumptions that capability implies robustness and that stronger feedback yields better outputs.
title RAudit: A Blind Auditing Protocol for Large Language Model Reasoning
topic Artificial Intelligence
I.2.7
url https://arxiv.org/abs/2601.23133