CoChat
Sign Up
Cited by 69Explainable Artificial Intelligence (XAI)Topic ModelingNatural Language Processing Techniques

The DOI matches its Crossref record; 7 of the 7 references it cites were found in a registry or an index.

Faithfulness Tests for Natural Language Explanations

Open access
Atanasova, Pepa, Camburu, Oana-Maria, Lioma, Christina, Lukasiewicz, Thomas, Simonsen, Jakob Grue, Augenstein, Isabelle·2023·Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)
DOI 10.18653/v1/2023.acl-short.25
Open this paper in CoChat — free accountOpen PDF

Abstract

Pepa Atanasova, Oana-Maria Camburu, Christina Lioma, Thomas Lukasiewicz, Jakob Grue Simonsen, Isabelle Augenstein. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers). 2023.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.18653/v1/2023.acl-short.25

Cites

7 of 7 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Recent research demonstrates that chains of thought are not entirely faithful to the model's underlying decision-making process.

    “Recent research demonstrates that chains of thought are not entirely faithful to the model's underlying decision-making process (Agarwal et al., 2024;Atanasova et al., 2023;Turpin et al., 2023;Lanham et al., 2023;Ye & Durrett, 2022).”

    From Open Problems in Mechanistic Interpretability (2025) · cites this paper as Atanasova et al., 2023

    The Summary and Outlook reports that all four tested setups of natural-language explanation models are prone to generate unfaithful explanations. This supports the broader point that explanations need not reflect a model’s decision-making process, but the paper does not specifically establish the claim for chains of thought. (LLM, conf=0.82)

    Experimental — accuracy not yet measured

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Flagged first, then by citations

Research works printed but not linked: 34

  1. Julius Adebayo, Michael Muelly, Harold Abelson, and Been Kim. 2022. Post hoc explanations may be in- effective for detecting unknown spurious correlation. In International Conference on Learning Represen- tations.
  2. Christopher Anders, Plamen Pasliev, Ann-Kathrin Dombrowski, Klaus-Robert Müller, and Pan Kessel. 2020. Fairwashing explanations with off-manifold detergent. In International Conference on Machine Learning, pages 314-323.
  3. PMLR.
  4. Pepa Atanasova, Jakob Grue Simonsen, Christina Li- oma, and Isabelle Augenstein. 2020a. A diagnos- tic study of explainability techniques for text clas- sification. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Process- ing (EMNLP), pages 3256-3274, Online. Associa- tion for Computational Linguistics. · DOI
  5. Pepa Atanasova, Jakob Grue Simonsen, Christina Li- oma, and Isabelle Augenstein. 2020b. Generating fact checking explanations. In Proceedings of the 58th Annual Meeting of the Association for Compu- tational Linguistics, pages 7352-7364, Online. As- sociation for Computational Linguistics. · DOI
  6. Samuel R. Bowman, Gabor Angeli, Christopher Potts, and Christopher D. Manning. 2015. A large anno- tated corpus for learning natural language inference. In Proceedings of the 2015 Conference on Empiri- cal Methods in Natural Language Processing, pages 632-642, Lisbon, Portugal. Association for Compu- tational Linguistics. · DOI · Paper page
  7. Oana-Maria Camburu, Eleonora Giunchiglia, Jakob Foerster, Thomas Lukasiewicz, and Phil Blunsom. 2019. Can I Trust the Explainer? Verifying Post-hoc Explanatory Methods. In NeurIPS 2019 Workshop Safety and Robustness in Decision Making.
  8. Oana-Maria Camburu, Eleonora Giunchiglia, Jakob Foerster, Thomas Lukasiewicz, and Phil Blunsom. 2021. The struggles of feature-based explanations: Shapley values vs. minimal sufficient subsets. In AAAI 2021 Workshop on Explainable Agency in Ar- tificial Intelligence.
  9. Oana-Maria Camburu, Brendan Shillingford, Pasquale Minervini, Thomas Lukasiewicz, and Phil Blunsom. 2020. Make up your mind! adversarial generation of inconsistent natural language explanations. In Proceedings of the 58th Annual Meeting of the Asso- ciation for Computational Linguistics, pages 4157- 4165, Online. Association for Computational Lin- guistics. · DOI
  10. Aaron Chan, Shaoliang Nie, Liang Tan, Xiaochang Peng, Hamed Firooz, Maziar Sanjabi, and Xiang Ren. 2022a. Frame: Evaluating simulatability metrics for free-text rationales. arXiv preprint arXiv:2207.00779. · arXiv
  11. Chun Sik Chan, Huanqi Kong, and Liang Guanqing. 2022b. A comparative study of faithfulness metrics for model interpretability methods. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 5029-5038, Dublin, Ireland. Association for Computational Linguistics. · DOI
  12. Christiane Fellbaum. 2010. Wordnet. In Theory and Applications of Ontology: Computer Applications, pages 231-243. Springer.
  13. Riccardo Guidotti. 2022. Counterfactual explanations and how to find them: literature review and bench- marking. Data Mining and Knowledge Discovery, pages 1-55.
  14. Leif Hancox-Li. 2020. Robustness in machine learning explanations: does it matter? In Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency, pages 640-647.
  15. Peter Hase, Shiyue Zhang, Harry Xie, and Mohit Bansal. 2020. Leakage-adjusted simulatability: Can models generate non-trivial explanations of their be- havior in natural language? In Findings of the As- sociation for Computational Linguistics: EMNLP 2020, pages 4351-4367, Online. Association for Computational Linguistics. · DOI · Paper page
  16. Cheng-Yu Hsieh, Chih-Kuan Yeh, Xuanqing Liu, Pradeep Kumar Ravikumar, Seungyeon Kim, Sanjiv Kumar, and Cho-Jui Hsieh. 2021. Evaluations and methods for explanation through robustness analysis. In International Conference on Learning Represen- tations.
  17. Alon Jacovi, Swabha Swayamdipta, Shauli Ravfogel, Yanai Elazar, Yejin Choi, and Yoav Goldberg. 2021. Contrastive explanations for model interpretability. In Proceedings of the 2021 Conference on Empiri- cal Methods in Natural Language Processing, pages 1597-1611, Online and Punta Cana, Dominican Re- public. Association for Computational Linguistics.
  18. Shailza Jolly, Pepa Atanasova, and Isabelle Augenstein. 2022. Generating Fluent Fact Checking Explana- tions with Unsupervised Post-Editing. Information, 13(10). · DOI
  19. Maxime Kayser, Cornelius Emde, Oana-Maria Cam- buru, Guy Parsons, Bartlomiej Papiez, and Thomas Lukasiewicz. 2022. Explaining chest x-ray patholo- gies in natural language. In Medical Image Com- puting and Computer Assisted Intervention -MIC- CAI 2022, pages 701-713, Cham. Springer Nature Switzerland.
  20. Andreas Madsen, Nicholas Meade, Vaibhav Adlakha, and Siva Reddy. 2022. Evaluating the Faithful- ness of Importance Measures in NLP by Recursively Masking Allegedly Important Tokens and Retrain- ing. In Findings of the Association for Computa- tional Linguistics: EMNLP 2022, pages 1731-1751, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
  21. Bodhisattwa Prasad Majumder, Oana Camburu, Thomas Lukasiewicz, and Julian Mcauley. 2022. Knowledge-grounded self-rationalization via ex- tractive and natural language explanations. In Proceedings of the 39th International Conference on Machine Learning, volume 162 of Proceedings of Machine Learning Research, pages 14786-14801.
  22. PMLR.
  23. Ana Marasović, Iz Beltagy, Doug Downey, and Matthew E. Peters. 2022. Few-shot self- rationalization with natural language prompts. Findings of NAACL.
  24. Nazneen Fatema Rajani, Bryan McCann, Caiming Xiong, and Richard Socher. 2019. Explain yourself! leveraging language models for commonsense rea- soning. In Proceedings of the 57th Annual Meet- ing of the Association for Computational Linguis- tics, pages 4932-4942, Florence, Italy. Association for Computational Linguistics. · DOI
  25. Alexis Ross, Ana Marasović, and Matthew Peters. 2021. Explaining NLP models via minimal con- trastive editing (MiCE). In Findings of the Associ- ation for Computational Linguistics: ACL-IJCNLP 2021, pages 3840-3852, Online. Association for Computational Linguistics. · DOI
  26. Dylan Slack, Anna Hilgard, Himabindu Lakkaraju, and Sameer Singh. 2021. Counterfactual explanations can be manipulated. Advances in Neural Informa- tion Processing Systems, 34:62-75.
  27. Dylan Slack, Sophie Hilgard, Emily Jia, Sameer Singh, and Himabindu Lakkaraju. 2020. Fooling lime and shap: Adversarial attacks on post hoc explanation methods. In Proceedings of the AAAI/ACM Con- ference on AI, Ethics, and Society, AIES '20, page 180-186, New York, NY, USA. Association for Computing Machinery. · DOI · Paper page
  28. Jiao Sun, Swabha Swayamdipta, Jonathan May, and Xuezhe Ma. 2022. Investigating the benefits of free- form rationales. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 5867-5882, Abu Dhabi, United Arab Emirates. As- sociation for Computational Linguistics.
  29. Cunxiang Wang, Shuailong Liang, Yili Jin, Yi- long Wang, Xiaodan Zhu, and Yue Zhang. 2020. SemEval-2020 task 4: Commonsense validation and explanation. In Proceedings of the Four- teenth Workshop on Semantic Evaluation, pages 307-321, Barcelona (online). International Commit- tee for Computational Linguistics. · DOI
  30. Sarah Wiegreffe and Ana Marasovic. 2021. Teach Me to Explain: A Review of Datasets for Explainable Natural Language Processing. In Thirty-fifth Con- ference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 1).
  31. Tongshuang Wu, Marco Tulio Ribeiro, Jeffrey Heer, and Daniel Weld. 2021. Polyjuice: Generating coun- terfactuals for explaining, evaluating, and improving models. In Proceedings of the 59th Annual Meet- ing of the Association for Computational Linguistics and the 11th International Joint Conference on Nat- ural Language Processing (Volume 1: Long Papers), pages 6707-6723, Online. Association for Computa- tional Linguistics. · DOI
  32. Fan Yin, Zhouxing Shi, Cho-Jui Hsieh, and Kai-Wei Chang. 2022. On the Sensitivity and Stability of Model Interpretations in NLP. In Proceedings of the 60th Annual Meeting of the Association for Compu- tational Linguistics (Volume 1: Long Papers), pages 2631-2647, Dublin, Ireland. Association for Com- putational Linguistics. · DOI
  33. Yordan Yordanov, Vid Kocijan, Thomas Lukasiewicz, and Oana-Maria Camburu. 2022. Few-Shot Out-of- Domain Transfer of Natural Language Explanations. In Proceedings of the Findings of the Conference on Empirical Methods in Natural Language Processing (EMNLP).
  34. Mo Yu, Shiyu Chang, Yang Zhang, and Tommi Jaakkola. 2019. Rethinking cooperative rationaliza- tion: Introspective extraction and complement con- trol. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natu- ral Language Processing (EMNLP-IJCNLP), pages 4094-4103, Hong Kong, China. Association for Computational Linguistics. Model Acc↑ BLEU↑ SNLI MT-Re 88.24 20.01 ST-Re 87.68 19.67 MT-Ra 88.10 20.67 ST-Ra 87.63 20.59 CoS-E MT-Re 65.79 5.75 ST-Re 66.11 6.66 MT-Ra 66.95 5.55 ST-Ra 67.79 7.85 ComVE MT-Re 85.70 7.53 ST-Re 84.40 6.68 MT-Ra 86.40 7.03 ST-Ra 86.40 7.21 · DOI

Cited by

Sorted by citations

Metadata

Authors
Atanasova, P., Camburu, O.-M., Lioma, C., Lukasiewicz, T., Simonsen, J. G., & Augenstein, I.
Affiliations
  • University of Copenhagen (DK)
  • University College London (GB)
  • TU Wien (AT)
ORCID
Venue
Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)
Publisher
Association for Computational Linguistics
Published
2023
Pages
283-294
ISSN
0736-587X (ISSN-L)
DOI
10.18653/v1/2023.acl-short.25
OpenAlex
W4385571986
Cited by
69
References
7
Citations by year
2026: 31 · 2025: 18 · 2024: 14 · 2023: 6
FWCI
8.00
Type
conference paper
Language
en
Open access
Licence
cc-by
Locations
  • Proceedings of the conference - Association for Computational Linguistics. Meeting (publishedVersion, cc-by)
  • Research at the University of Copenhagen (University of Copenhagen) (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion, cc-by)
  • UCL Discovery (University College London) (submittedVersion, cc-by-nc-sa)
  • arXiv (Cornell University) (cc-by)
Funders
  • Alan Turing Institute (EP/N510129/1., N510129)
  • Leverhulme Trust
  • European Commission (801199, HORIZON2020, 952215)
  • Danmarks Grundforskningsfond
  • AXA Research Fund
  • Engineering and Physical Sciences Research Council (N510129, EP/N510129/, EP/N510129/1, EPSRC grant EP/N510129/1)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
  • Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
  • Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
Keywords
faithfulness, natural language explanations, explanation evaluation, explainable AI

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review