mlsanityData to 14 Oct 2026

A Necessary Step toward Faithfulness: Measuring and Improving Consistency in Free-Text Explanations

arXiv 2505.19299 ↗First posted 25 May 2025cs.CL cs.AIcited by 3

Not on the map. Why?

Abstract (arXiv)

Faithful free-text explanations are important to ensure transparency in high-stakes AI decision-making contexts, but they are challenging to generate by language models and assess by humans. In this paper, we present a measure for Prediction-EXplanation (PEX) consistency, by extending the concept of weight of evidence. This measure quantifies how much a free-text explanation supports or opposes a prediction, serving as an important aspect of explanation faithfulness. Our analysis reveals that more than 62% explanations generated by large language models lack this consistency. We show that applying direct preference optimization improves the consistency of generated explanations across three model families, with improvement ranging from 43.1% to 292.3%. Furthermore, we demonstrate that optimizing this consistency measure can improve explanation faithfulness by up to 9.7%.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 39 of its 51 references that are in the map. First posted on arXiv 25 May 2025.

Builds on

39 of 51 references.

References · 39

  1. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  2. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  4. 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,733
  5. 2017A Unified Approach to Interpreting Model Predictionscited by 6,555
  6. 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,070
  7. 2017Towards A Rigorous Science of Interpretable Machine Learningcited by 1,416
  8. 2017Explanation in Artificial Intelligence: Insights from the Social Sciencescited by 1,206
  9. 2023Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Promptingcited by 1,122
  10. 2024Yi: Open Foundation Models by 01.AIcited by 719
  11. 2017Program Induction by Rationale Generation : Learning to Solve and Explain Algebraic Word Problemscited by 650
  12. 2020Towards Faithfully Interpretable NLP Systems: How should we define and evaluate faithfulness?cited by 599
27 more
  1. 2018e-SNLI: Natural Language Inference with Natural Language Explanationscited by 468
  2. 2015Evaluating the visualization of what a Deep Neural Network has learnedcited by 419
  3. 2019Explain Yourself! Leveraging Language Models for Commonsense Reasoningcited by 367
  4. 2018On the Robustness of Interpretability Methodscited by 314
  5. 2018Learning to Explain: An Information-Theoretic Perspective on Model Interpretationcited by 248
  6. 2020Evaluating Explainable AI: Which Algorithmic Explanations Help Users Predict Model Behavior?cited by 190
  7. 2018Multimodal Explanations: Justifying Decisions and Pointing to the Evidencecited by 179
  8. 2022The Unreliability of Explanations in Few-shot Prompting for Textual Reasoningcited by 169
  9. 2022Maieutic Prompting: Logically Consistent Reasoning with Recursive Explanationscited by 159
  10. 2020WT5?! Training Text-to-Text Models to Explain their Predictionscited by 148
  11. 2022Towards Faithful Model Explanation in NLP: A Surveycited by 130
  12. 2021Reframing Human-AI Collaboration for Generating Free-Text Explanationscited by 125
  13. 2020NILE : Natural Language Inference with Faithful Natural Language Explanationscited by 96
  14. 2020Leakage-Adjusted Simulatability: Can Models Generate Non-Trivial Explanations of Their Behavior in Natural Language?cited by 95
  15. 2020Evaluating Explanations: How much do explanations from the teacher aid students?cited by 94
  16. 2021Few-Shot Self-Rationalization with Natural Language Promptscited by 83
  17. 2019Make Up Your Mind! Adversarial Generation of Inconsistent Natural Language Explanationscited by 71
  18. 2018Faithful Multimodal Explanation for Visual Question Answeringcited by 64
  19. 2024Towards Consistent Natural-Language Explanations via Explanation-Consistency Finetuningcited by 26
  20. 2023Tailoring Self-Rationalizers with Multi-Reward Distillationcited by 21
  21. 2018Interpreting Neural Networks With Nearest Neighborscited by 18
  22. 2021From Human Explanation to Model Interpretability: A Framework Based on Weight of Evidencecited by 17
  23. 2023FLamE: Few-shot Learning from Natural Language Explanationscited by 10
  24. 2023Explaining with Contrastive Phrasal Highlighting: A Case Study in Assisting Humans to Detect Translation Differencescited by 8
  25. 2023Is Ignorance Bliss? The Role of Post Hoc Explanation Faithfulness and Alignment in Model Trust in Laypeople and Domain Expertscited by 7
  26. 2024Fool Me Once? Contrasting Textual and Visual Explanations in a Clinical Decision-Support Settingcited by 5
  27. 2024ACORN: Aspect-wise Commonsense Reasoning Explanation Evaluationcited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 2; 2026: 1.

3 papers.

  1. 2025Argument-Based Consistency in Toxicity Explanations of LLMscited by 2
  2. 2025Support-Contra Asymmetry in LLM Explanationscited by 1
  3. 2026FRACTION: Fraud Explanation Generation via Agentic Graph Context Navigationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024The Probabilities Also Matter: A More Faithful Metric for Faithfulness of Free-Text Explanations in Large Language Models
  2. 2026ICE: Intervention-Consistent Explanation Evaluation with Statistical Grounding for LLMs
  3. 2023On Measuring Faithfulness or Self-consistency of Natural Language Explanations
  4. 2024Faithfulness vs. Plausibility: On the (Un)Reliability of Explanations from Large Language Models
  5. 2025A Causal Lens for Evaluating Faithfulness Metrics
  6. 2026Towards Faithful Agentic XAI: A Verification Method and an Open-World Benchmark for Better Model Faithfulness
  7. 2024Towards Faithful Natural Language Explanations: A Study Using Activation Patching in Large Language Models
  8. 2023Benchmarking Faithfulness: Towards Accurate Natural Language Explanations in Vision-Language Tasks