mlsanityData to 14 Oct 2026

MetaFaith: Faithful Natural Language Uncertainty Expression in LLMs

arXiv 2505.24858 ↗First posted 30 May 2025cs.CL cs.LGcited by 16

Not on the map. Why?

Abstract (arXiv)

A critical component in the trustworthiness of LLMs is reliable uncertainty communication, yet LLMs often use assertive language when conveying false claims, leading to over-reliance and eroded trust. We present the first systematic study of textitfaithful confidence calibration of LLMs, benchmarking models' ability to use linguistic expressions of uncertainty that textitfaithfully reflect their intrinsic uncertainty, across a comprehensive array of models, datasets, and prompting strategies. Our results demonstrate that LLMs largely fail at this task, and that existing interventions are insufficient: standard prompt approaches provide only marginal gains, and existing, factuality-based calibration techniques can even harm faithful calibration. To address this critical gap, we introduce MetaFaith, a novel prompt-based calibration approach inspired by human metacognition. We show that MetaFaith robustly improves faithful calibration across diverse models and task domains, enabling up to 61% improvement in faithfulness and achieving an 83% win rate over original generations as judged by humans.

Key facts. Cited by 16 papers in the map (data through 14 Oct 2026). Builds on 66 of its 118 references that are in the map. First posted on arXiv 30 May 2025.

Builds on

66 of 118 references.

References · 66

  1. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  2. 2017On Calibration of Modern Neural Networkscited by 4,960
  3. 2018GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understandingcited by 4,899
  4. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,886
  5. 2017TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehensioncited by 2,955
  6. 2019SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systemscited by 1,729
  7. 2023Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMscited by 974
  8. 2023Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedbackcited by 812
  9. 2022Discovering Latent Knowledge in Language Models Without Supervisioncited by 767
  10. 2017Crowdsourcing Multiple Choice Science Questionscited by 765
  11. 2022Teaching Models to Express Their Uncertainty in Wordscited by 751
  12. 2023The Internal State of an LLM Knows When It's Lyingcited by 659
38 more
  1. 2020Towards Faithfully Interpretable NLP Systems: How should we define and evaluate faithfulness?cited by 599
  2. 2020How Can We Know When Language Models Know? On the Calibration of Language Models for Question Answeringcited by 445
  3. 2024Measuring short-form factuality in large language modelscited by 373
  4. 2020Calibration of Pre-trained Transformerscited by 315
  5. 2020Effect of Confidence and Explanation on Accuracy and Trust Calibration in AI-Assisted Decision Makingcited by 281
  6. 2022Prompting GPT-3 To Be Reliablecited by 260
  7. 2023A Survey of Confidence Estimation and Calibration in Large Language Modelscited by 256
  8. 2023Do Large Language Models Know What They Don't Know?cited by 243
  9. 2020Selective Question Answering under Domain Shiftcited by 238
  10. 2024Patchscopes: A Unifying Framework for Inspecting Hidden Representations of Language Modelscited by 235
  11. 2024Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?cited by 232
  12. 2023Fine-tuning Language Models for Factualitycited by 221
  13. 2020Reducing conversational agents' overconfidence through linguistic calibrationcited by 210
  14. 2021On Hallucination and Predictive Uncertainty in Conditional Language Generationcited by 176
  15. 2024Kernel Language Entropy: Fine-grained Uncertainty Quantification for LLMs from Semantic Similaritiescited by 159
  16. 2023LM-Polygraph: Uncertainty Estimation for Language Modelscited by 131
  17. 2022Towards Faithful Model Explanation in NLP: A Surveycited by 130
  18. 2023Quantifying Uncertainty in Answers from any Language Model and Enhancing their Trustworthinesscited by 122
  19. 2024To Believe or Not to Believe Your LLMcited by 120
  20. 2023Decomposing Uncertainty for Large Language Models through Input Clarification Ensemblingcited by 114
  21. 2024What Large Language Models Know and What People Think They Knowcited by 110
  22. 2024Relying on the Unreliable: The Impact of Language Models' Reluctance to Express Uncertaintycited by 106
  23. 2022Uncertainty Quantification with Pre-trained Language Models: A Large-Scale Empirical Analysiscited by 104
  24. 2023Navigating the Grey Area: How Expressions of Uncertainty and Overconfidence Affect Language Modelscited by 91
  25. 2024Can Large Language Models Faithfully Express Their Intrinsic Uncertainty in Words?cited by 87
  26. 2024On Verbalized Confidence Scores for LLMscited by 83
  27. 2024SelfIE: Self-Interpretation of Large Language Model Embeddingscited by 67
  28. 2025A Survey of Uncertainty Estimation Methods on Large Language Modelscited by 62
  29. 2023Uncertainty in Natural Language Generation: From Theory to Applicationscited by 61
  30. 2024Linguistic Calibration of Long-Form Generationscited by 60
  31. 2023Metacognitive Prompting Improves Understanding in Large Language Modelscited by 58
  32. 2022A Close Look into the Calibration of Pre-trained Language Modelscited by 57
  33. 2024A Survey of Uncertainty Estimation in LLMs: Theory Meets Practicecited by 55
  34. 2025Calibrating Verbal Uncertainty as a Linear Feature to Reduce Hallucinationscited by 48
  35. 2023LitCab: Lightweight Language Model Calibration over Short- and Long-form Responsescited by 44
  36. 2024Benchmarking Hallucination in Large Language Models based on Unanswerable Math Word Problemcited by 43
  37. 2023Llamas Know What GPTs Don't Show: Surrogate Models for Confidence Estimationcited by 39
  38. 2024Confidence Under the Hood: An Investigation into the Confidence-Probability Alignment in Large Language Modelscited by 34

50 most cited of 66 All 66 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 3; 2026: 13.

16 papers.

Sort:

By area:

  1. 2025LLMs Don't Know Their Own Decision Boundaries: The Unreliability of Self-Generated Counterfactual Explanationscited by 16
  2. 2025Anthropomimetic Uncertainty: What Verbalized Uncertainty in Language Models is Missingcited by 14
  3. 2025Teaching Language Models to Faithfully Express their Uncertaintycited by 7
  4. 2026Hallucinations Undermine Trust; Metacognition is a Way Forwardcited by 5
  5. 2026Fine-Tuning Language Models to Know What They Knowcited by 2
  6. 2026Can LLMs Use Linguistic Uncertainty Markers to Reliably Reflect Intrinsic Confidence?cited by 2
  7. 2026Quantifying Faithful Confidence Expression in Large Reasoning Modelscited by 2
  8. 2026Belief-Guided Inference Control for Large Language Model Services via Verifiable Observationscited by 1
  9. 2026Metacognition in LLMs: Foundations, Progress, and Opportunitiescited by 1
  10. 2026An Information-Geometric Framework for Stability Analysis of Large Language Models under Entropic Stresscited by 0
  11. 2026Faithful by Definition: Emotion Analysis via Natural Semantic Metalanguage Explicationscited by 0
  12. 2026Evaluation Pitfalls and Sparsity Limitations in LLM-based Confidence Estimates for Classificationcited by 0
4 more
  1. 2026On the Role of Citations in Preference Datacited by 0
  2. 2026Can a Model Catch Its Own Hallucinations for Free?: Label-Free Doubt Signals Hold Their Own Against a Labelled Dataset for Abstentioncited by 0
  3. 2026Stochastic Semantic Evidence Graphs: Uncertainty Propagation and Governance for Agentic AIcited by 0
  4. 2026The Confidence Game: Strategic Miscalibration in Human-AI Delegationcited by 0