Metraipolis › archive › dc_5b81978d467c

Right Answer, Wrong Reason: Accuracy, Consistency, and Consensus Are Misleading Indicators of LLM Faithfulness in Clinical Decision Support — arXiv:2609.32817

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.32817
License (as archived)
CC0-1.0
Archived
2026-10-07T20:56:10.549Z
SHA-256 of the archived text
370fffd74baa1a6bd132c194ecb336c1c1f22ec64b100497cda6c4d5527fc836
Stable URL of this page
https://metraipolis.com/documento/dc_5b81978d467c

How to cite

Right Answer, Wrong Reason: Accuracy, Consistency, and Consensus Are Misleading Indicators of LLM Faithfulness in Clinical Decision Support — arXiv:2609.32817 — arXiv (en), https://arxiv.org/abs/2609.32817 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5b81978d467c (Metraipolis archive dc_5b81978d467c, sha256 370fffd74baa1a6bd132c194ecb336c1c1f22ec64b100497cda6c4d5527fc836)
BibTeX
@misc{dc5b81978d467c,
  title = {Right Answer, Wrong Reason: Accuracy, Consistency, and Consensus Are Misleading Indicators of LLM Faithfulness in Clinical Decision Support — arXiv:2609.32817},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.32817},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5b81978d467c; sha256 370fffd74baa1a6bd132c194ecb336c1c1f22ec64b100497cda6c4d5527fc836},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32817. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Clinical Large Language Models (LLMs) achieve strong medical-exam accuracy; however, a correct answer does not guarantee that the explanation names the concepts that actually drove the decision. We introduce three lightweight, directly interpretable metrics for this faithfulness gap: the Explanation Stability Index (ESI), which measures reasoning consistency across repeated queries; the Causal Faithfulness Score (CFS), which tests whether cited concepts drive predictions via concept ablation; and the Perturbation Stability Score (PSS), which measures robustness to semantic-preserving paraphrases. By evaluating six LLMs on 150 MedQA-USMLE questions (900 model-question observations), we found that only 23.3% of the cited clinical concepts were causally necessary. Correct answers had lower CFS than incorrect answers (0.212 vs. 0.398), answer consistency negatively predicted CFS (Spearman r = -0.466), and model pairs could agree on answers while sharing only 8.8% of cited reasoning concepts. These results show that accuracy, consistency, and consensus are incomplete safety signals for clinical decision-making support. The evidence is behavioral rather than mechanistic: concept ablation tests counterfactual sensitivity of outputs, not internal circuits.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.32817 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T20:29:01.005Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).