# When LLMs Sit Above Diagnostic Tools: Unrealized Complementarity in Industrial Fault Diagnosis

arXiv: https://arxiv.org/abs/2610.05031 · HTML: https://mlsanity.com/paper/2610.05031/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 24 references that are in the map.
- First posted on arXiv 4 Oct 2026.

## Builds on (most cited first, 19 of 19)

- [Towards Understanding Sycophancy in Language Models](https://mlsanity.com/paper/2310.13548/) (2023)
- [Discovering Language Model Behaviors with Model-Written Evaluations](https://mlsanity.com/paper/2212.09251/) (2022)
- [Adaptive Chameleon or Stubborn Sloth: Revealing the Behavior of Large Language Models in Knowledge Conflicts](https://mlsanity.com/paper/2305.13300/) (2023)
- [Consistent Estimators for Learning to Defer to an Expert](https://mlsanity.com/paper/2006.01862/) (2020)
- [Predict Responsibly: Improving Fairness and Accuracy by Learning to Defer](https://mlsanity.com/paper/1711.06664/) (2017)
- [When combinations of humans and AI are useful: A systematic review and meta-analysis](https://mlsanity.com/paper/2405.06087/) (2024)
- [Appropriate Reliance on AI Advice: Conceptualization and the Effect of Explanations](https://mlsanity.com/paper/2302.02187/) (2023)
- [Differentiable Learning Under Triage](https://mlsanity.com/paper/2103.08902/) (2021)
- [Who Should Predict? Exact Algorithms For Learning to Defer to Humans](https://mlsanity.com/paper/2301.06197/) (2023)
- [Whose Facts Win? LLM Source Preferences under Knowledge Conflicts](https://mlsanity.com/paper/2601.03746/) (2026)
- [Are Diversity Metrics Measuring Diversity? A Capability-Controlled Audit of Majority-Vote Gain in LLM Ensembles](https://mlsanity.com/paper/2607.20768/) (2026)
- [Evidence-Driven Reasoning for Industrial Maintenance Using Heterogeneous Data](https://mlsanity.com/paper/2603.08171/) (2026)
- [When to Trust Tools? Adaptive Tool Trust Calibration For Tool-Integrated Math Reasoning](https://mlsanity.com/paper/2604.08281/) (2026)
- [Authority Bias in Language Models: Source Deference and User Agreement Are Not Interchangeable](https://mlsanity.com/paper/2609.37616/) (2026)
- [FaultExplainer: Leveraging Large Language Models for Interpretable Fault Detection and Diagnosis](https://mlsanity.com/paper/2412.14492/) (2024)
- [S2S-FDD: Bridging Industrial Time Series and Natural Language for Explainable Zero-shot Fault Diagnosis](https://mlsanity.com/paper/2603.08048/) (2026)
- [TrustMargin: Training-Free Arbitration between Parametric Memory and Retrieved Evidence in Large Language Models](https://mlsanity.com/paper/2606.08397/) (2026)
- [Large Language Models in Process Systems Engineering: Opportunities, Architectures, and Industrial Deployment Challenges](https://mlsanity.com/paper/2606.11589/) (2026)
- [MemToC: Benchmarking Memory-Tool Conflict Resolution in Large Language Models](https://mlsanity.com/paper/2608.26295/) (2026)

## Built on by (most cited first, 1 of 1)

- [Where Do LLM Ensemble Gains Come From? Decomposing Candidate Value, Diversity, and Aggregation](https://mlsanity.com/paper/2610.18786/) (2026)

## Abstract (arXiv)

Large language models are increasingly used as integration layers above specialized tools, but a stronger component does not necessarily produce a stronger combined system. Across five diagnostic datasets (bearing vibration, process monitoring, semiconductor equipment), we study whether an LLM can reliably use external diagnostic information; paired repeat calls separate advice effects from output instability. In all five, conflicting external information overturned initially correct LLM judgments. Among the four datasets with direct integration comparisons, none showed a consistent advantage for implicit LLM integration over the stronger standalone source. On a Tennessee Eastman confirmation set whose protocol was fixed before evaluation, unaided accuracy was 64.67%, implicit LLM-specialist integration 77.43%, and the specialist alone 83.33%. Specialist information improved the LLM by 12.8 points (95% interval 9.7 to 15.9), yet the integrated output stayed 5.9 points below the specialist (95% interval -12.0 to -0.7). A two-source selector oracle reached 92.76%, indicating complementarity that the integrated output did not fully realize. The integrated output missed 140 of 295 specialist corrections (47.5%) but lost 15 of 99 initially correct LLM judgments (15.2%). The deficit remained under prompt and specialist sensitivity analyses. Among CWRU cases solved under both evidence presentations, task-aligned physical evidence yielded lower estimates of susceptibility to incorrect advice in six of seven models (five intervals excluding zero); higher reasoning effort gave no reliable reduction in five models, and a separate four-model TEP analysis gave no clear evidence that it resolves the integration problem. Source quality and integration quality should be evaluated separately: an integration layer should be compared with its stronger standalone component, not only with the unaided LLM.