When Derived Measurements Mislead: Quantifying and Mitigating LLM Over-Trust with Privileged-Modality Reliability Evidence
Derived measurements increasingly enter large language model (LLM) pipelines as direct facts despite their instance-dependent validity. We define derived-feature over-trust (DFOT) as the failure in which a downstream LLM assigns such a measurement the epistemic status of a direct fact or uses it outside its valid scope. Using physiological sensing as a case study, D1 tests acceptance of a PPG-derived rhythm contradicted by offline ECG, whereas D2 tests rejection of an offline-confirmed reliable PPG rhythm under misleading severe history. ECG supplies training supervision and offline reference construction but is never shown to the LLM. Five estimands quantify this chain: conflict over-trust rate (COTR) and context-induced error rate (CIR) characterize D1/D2; correct repair rate (CRR) measures frozen-error repair; evidence-specific repair margin (ESRM) contrasts matched and patient-disjoint shuffled evidence; and utility harm rate (UHR) measures unnecessary verification among HIGH-reliability cases used without verification at baseline. The framework does not depend on a particular reliability generator. We demonstrate it on 50,000 paired PPG-ECG records using ECG-to-PPG privileged distillation as an illustrative baseline and PPG-only inference. On a protocol-locked 187-patient test, the baseline improves four repair and specificity endpoints by 1.82-6.69 percentage points, with all paired confidence intervals excluding zero; UHR increases by 0.67 percentage points (95% CI: -0.4 to +1.7). DFOT provides a common evaluation target for stronger mitigation methods. The code is available at https://github.com/Zongheng-Guo/When-Derived-Measurements-Mislead.
Code (1)
Similar Papers 제목 키워드 기반
Quantifying and Mitigating Self-Preference Bias of LLM Judges
LLM-as-a-Judge has become a dominant approach in automated evaluation systems, playing critical roles in model alignment, leaderboard construction, quality control, and so on. However, the scalability and trustworthiness…
A Particle Filtering Framework for Integrity Risk of GNSS-Camera Sensor Fusion
Adopting a joint approach towards state estimation and integrity monitoring results in unbiased integrity monitoring unlike traditional approaches. So far, a joint approach was used in Particle RAIM [l] for GNSS measurem…
PositionSensor FusionState EstimationThe Impact of Unstated Norms in Bias Analysis of Language Models
Bias in large language models (LLMs) has many forms, from overt discrimination to implicit stereotypes. Counterfactual bias evaluation is a widely used approach to quantifying bias and often relies on template-based prob…
Bias DetectioncounterfactualSentiment AnalysisID-XCB: Data-independent Debiasing for Fair and Accurate Transformer-based Cyberbullying Detection
Swear words are a common proxy to collect datasets with cyberbullying incidents. Our focus is on measuring and mitigating biases derived from spurious associations between swear words and incidents occurring as a result …
Worse than Zero-shot? A Fact-Checking Dataset for Evaluating the Robustness of RAG Against Misleading Retrievals
Retrieval-augmented generation (RAG) has shown impressive capabilities in mitigating hallucinations in large language models (LLMs). However, LLMs struggle to handle misleading retrievals and often fail to maintain their…
Fact CheckingMisinformationNavigateRAG+2