pith. sign in

arxiv: 2606.07101 · v1 · pith:X5X33QFAnew · submitted 2026-06-05 · 💻 cs.HC

CANote: Empowering Fact-checking Note Writing Through Scaffolded and Provenance-based Human-AI Collaboration

Shuning Zhang , Jingruo Chen , Yuwei Chuai , Dai Shi , Yifan Wang , Xin Yi , Hewu Li This is my paper

Pith reviewed 2026-06-27 21:02 UTC · model grok-4.3

classification 💻 cs.HC
keywords fact-checkinghuman-AI collaborationCommunity Notesmisinformation mitigationAI-assisted writingprovenancescaffolded workflowcrowdsourced debunking
0
0 comments X

The pith

CANote's AI scaffolding with subclaim extraction and provenance links lets lay users produce fact-check notes matching expert quality.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents CANote as a system that extracts subclaims from posts, links them explicitly to retrieved evidence, and supplies neutral structural drafts to guide human writers. This scaffolded approach is tested on a simulated X platform against unaided writing by 52 fact-checkers and 52 lay users. The central finding is that the system raises overall note quality while enabling lay users to reach the same standard as experts, without raising task time or cognitive load and while increasing satisfaction. A sympathetic reader would care because crowdsourced debunking currently burdens contributors, and lowering that barrier could expand participation in fighting misinformation. The work also records a side effect of lower user ownership over the final note.

Core claim

CANote improves note quality through evidence correlation and structured co-drafting by breaking posts into subclaims, supplying explicit provenance links to supporting evidence, and generating neutral structural drafts that support rather than replace human reasoning; in a controlled evaluation on a simulated X platform, both expert and lay users produced higher-quality notes with the system than without it, with lay users reaching quality levels comparable to unaided experts.

What carries the argument

Scaffolded workflow that extracts subclaims, provides explicit provenance links between subclaims and evidence, and generates neutral structural drafts for human co-drafting.

If this is right

  • Lay users become viable contributors of high-quality debunking notes.
  • Overall quality of crowdsourced notes on platforms like X rises.
  • Task completion time and cognitive load stay the same as manual writing.
  • User satisfaction with the writing process increases.
  • Contributors report a lower sense of ownership and control over the finished note.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same scaffolding pattern could be adapted to other social platforms that rely on user-generated corrections.
  • Lower ownership might reduce long-term contributor retention even if short-term quality improves.
  • The provenance links could serve as training data for fully automated fact-checking systems.
  • Real-world rollout would need to test whether the quality gains persist when users know the audience is live rather than simulated.

Load-bearing premise

The controlled simulation of the X platform with 52 participants per group accurately captures real-world note quality and generalizes to actual deployment.

What would settle it

In a live X deployment, notes written with CANote receive no higher quality ratings from platform users or independent raters than notes written without the system.

Figures

Figures reproduced from arXiv: 2606.07101 by Dai Shi, Hewu Li, Jingruo Chen, Shuning Zhang, Xin Yi, Yifan Wang, Yuwei Chuai.

Figure 1
Figure 1. Figure 1: The CANote workflow, which features a four-stage human-in-the-loop pipeline for AI-assisted debunking note [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The interface of CANote. In the first stage ( [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: System flow of CANote, illustrating the iterative human-in-the-loop pipeline. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Experimental interface of the baseline technique. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Toxicity and hateful-speech probabilities of notes [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 5
Figure 5. Figure 5: Note quality (left), note writing time (middle) and [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: NASA-TLX ratings across techniques (0=best per [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: User experience ratings across techniques (1=most [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Different rating dimensions across different tech [PITH_FULL_IMAGE:figures/full_fig_p008_9.png] view at source ↗
read the original abstract

Crowdsourced fact-checking mechanisms, such as X's Community Notes, play a critical role in mitigating the spread of misinformation. However, drafting high-quality, evidence-based debunking notes imposes a substantial burden on contributors. We present CANote, an AI-assisted debunking note writing system featuring evidence correlation and structured co-drafting. CANote scaffolds the workflow by extracting subclaims from social media posts, providing provenance through explicit links between subclaims and retrieved evidence, and generating neutral, structural drafts to support human reasoning. We evaluated CANote against manual writing (N=52 fact-checkers, N=52 lay users) on simulated X platform, where we found CANote significantly improves note quality. Notably, CANote enables lay users to write notes that have comparable quality to those written by experts. While the task completion time and perceived cognitive load remain comparable to manual drafting, CANote significantly increases user satisfaction. However, this assistance introduces a trade-off, resulting in a reduced sense of user ownership and control over the debunking note.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper presents CANote, an AI-assisted system for drafting evidence-based fact-checking notes on platforms like X's Community Notes. The system extracts subclaims from posts, links them explicitly to retrieved evidence for provenance, and generates neutral structural drafts to scaffold human reasoning. A controlled user study on a simulated X platform (N=52 fact-checkers and N=52 lay users) is reported to show that CANote significantly improves note quality over manual writing, enables lay users to reach expert-comparable quality, increases satisfaction, keeps task time and cognitive load similar, but reduces perceived ownership and control.

Significance. If the evaluation holds under scrutiny, the work could meaningfully advance human-AI collaboration tools for crowdsourced fact-checking by lowering barriers for non-experts while preserving evidence linkage. The provenance and scaffolding features directly target documented burdens in Community Notes-style systems, and the lay-expert parity result, if substantiated, would be a notable finding for participatory misinformation mitigation.

major comments (2)
  1. [Abstract / Evaluation] Abstract and user-study description: the claims of 'significantly improves note quality' and 'lay users to write notes that have comparable quality to those written by experts' are presented without any information on the quality rubric (dimensions scored, scoring scale), statistical tests used, blinding of raters to condition or author expertise level, or inter-rater reliability metrics. These omissions make it impossible to evaluate whether the reported significance and parity are interpretable or robust.
  2. [Abstract / Evaluation] Abstract and simulated-platform evaluation: the headline lay-expert comparability result rests entirely on the N=52/group controlled simulation, yet no details are supplied on how note visibility, feedback loops, or platform dynamics in the simulation map to real Community Notes mechanics, nor on any controls for content domain or post difficulty. This leaves the generalizability of the central claim unassessable.
minor comments (1)
  1. [Abstract] The abstract states that task completion time and cognitive load 'remain comparable' while satisfaction increases; a table or figure reporting the exact means, standard deviations, and p-values for these measures would improve clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for highlighting critical gaps in the reporting of our evaluation. We agree that the current manuscript version lacks sufficient detail on the quality assessment protocol and simulation design. We will revise the paper to include these elements, which will strengthen the interpretability of our results.

read point-by-point responses

  1. Referee: [Abstract / Evaluation] Abstract and user-study description: the claims of 'significantly improves note quality' and 'lay users to write notes that have comparable quality to those written by experts' are presented without any information on the quality rubric (dimensions scored, scoring scale), statistical tests used, blinding of raters to condition or author expertise level, or inter-rater reliability metrics. These omissions make it impossible to evaluate whether the reported significance and parity are interpretable or robust.

    Authors: We agree that the abstract and user-study description omit essential methodological details required to assess the robustness of the quality and parity claims. In the revised manuscript we will expand the Evaluation section to specify: (1) the full quality rubric with all scored dimensions and the 1-5 Likert scale; (2) the exact statistical tests (including effect sizes and corrections for multiple comparisons); (3) that raters were blinded to both experimental condition and whether the note author was a fact-checker or lay user; and (4) inter-rater reliability statistics (e.g., Krippendorff’s alpha or ICC). These additions will be placed in both the main text and a new supplementary table. revision: yes

  2. Referee: [Abstract / Evaluation] Abstract and simulated-platform evaluation: the headline lay-expert comparability result rests entirely on the N=52/group controlled simulation, yet no details are supplied on how note visibility, feedback loops, or platform dynamics in the simulation map to real Community Notes mechanics, nor on any controls for content domain or post difficulty. This leaves the generalizability of the central claim unassessable.

    Authors: We concur that the current description of the simulated platform is insufficient to evaluate ecological validity. In revision we will add a dedicated subsection detailing: (a) how note visibility and rating mechanics were approximated (including the absence of real-time community feedback loops, which we will explicitly flag as a limitation); (b) the procedure used to select and balance posts across content domains and difficulty levels; and (c) any pilot testing performed to calibrate the simulation. We will also add a limitations paragraph acknowledging that controlled simulations cannot fully replicate emergent platform dynamics such as iterative note editing by multiple contributors. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical user study with no derivations or fitted predictions

full rationale

The paper presents a system description and controlled user study (N=52 per group) comparing CANote-assisted note writing to manual drafting on a simulated platform. No equations, first-principles derivations, parameter fitting, or predictions appear in the abstract or described content. Claims of quality improvement and lay-expert parity rest on direct empirical measurement rather than any self-referential reduction or self-citation chain. The evaluation is self-contained against its own rubric and participant data; no load-bearing step reduces to an input by construction. This is the expected outcome for an HCI system paper.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The contribution rests on the domain premise that crowdsourced notes are valuable and that quality can be improved through structured AI support; no free parameters or new physical entities are introduced.

axioms (1)
  • domain assumption Crowdsourced fact-checking mechanisms play a critical role in mitigating the spread of misinformation.
    Opening sentence of the abstract states this as background motivation.
invented entities (1)
  • CANote system no independent evidence
    purpose: AI-assisted debunking note writing with evidence correlation and structured co-drafting
    The tool itself is the primary contribution of the work.

pith-pipeline@v0.9.1-grok · 5732 in / 1199 out tokens · 19548 ms · 2026-06-27T21:02:00.660024+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

99 extracted references · 1 linked inside Pith

  1. [1]

    Julius Adebayo, Justin Gilmer, Michael Muelly, Ian Goodfellow, Moritz Hardt, and Been Kim. 2018. Sanity checks for saliency maps.Advances in neural information processing systems31 (2018)

    2018

  2. [2]

    Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan, and Ameet Deshpande. 2024. Geo: Generative engine optimization. InProceedings of the 30th ACM SIGKDD conference on knowledge discovery and data mining. 5–16

    2024

  3. [3]

    Mubashara Akhtar, Michael Schlichtkrull, Zhijiang Guo, Oana Cocarascu, Elena Simperl, and Andreas Vlachos. 2023. Multimodal automated fact-checking: A survey. InFindings of the Association for Computational Linguistics: EMNLP 2023. 5430–5448

    2023

  4. [4]

    Mubashara Akhtar, Michael Schlichtkrull, and Andreas Vlachos. 2024. Ev2r: Evaluating evidence retrieval in automated fact-checking.arXiv preprint arXiv:2411.05375(2024)

    arXiv 2024

  5. [5]

    Maryam Ashoori and Justin D Weisz. 2019. In AI we trust? Factors that influ- ence trustworthiness of AI-infused decision-making processes.arXiv preprint arXiv:1912.02675(2019)

    arXiv 2019

  6. [6]

    Pepa Atanasova. 2024. Multi-hop fact checking of political claims. InAccountable and Explainable Methods for Complex Reasoning over Text. Springer, 131–151

    2024

  7. [7]

    Pepa Atanasova, Jakob Grue Simonsen, Christina Lioma, and Isabelle Augenstein

  8. [8]

    InProceedings of the 58th annual meeting of the association for computational linguistics

    Generating fact checking explanations. InProceedings of the 58th annual meeting of the association for computational linguistics. 7352–7364

  9. [9]

    Michael Bailey, David Dittrich, Erin Kenneally, and Doug Maughan. 2012. The menlo report.IEEE Security & Privacy10, 2 (2012), 71–75

    2012

  10. [10]

    Eivind Morris Bakke and Nora Winger Heggelund. 2025. (Fact) Check Your Bias. InProceedings of the Eighth Fact Extraction and VERification Workshop (FEVER). 162–178

    2025

  11. [11]

    Ramy Baly, Georgi Karadzhov, Dimitar Alexandrov, James Glass, and Preslav Nakov. 2018. Predicting factuality of reporting and bias of news media sources. InProceedings of the 2018 conference on empirical methods in natural language processing. 3528–3539

    2018

  12. [12]

    Mariano Barone, Antonio Romano, Giuseppe Riccio, Marco Postiglione, and Vincenzo Moscato. 2025. Combining Evidence and Reasoning for Biomedical Fact-Checking. InProceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval. 1087–1097

    2025

  13. [13]

    Azadeh Bayani, Alexandre Ayotte, Jean Noel Nikiema, et al. 2025. Transformer- based tool for automated fact-checking of online health information: Develop- ment study.JMIR infodemiology5, 1 (2025), e56831

    2025

  14. [14]

    Tom L Beauchamp et al. 2008. The belmont report.The Oxford textbook of clinical research ethics(2008), 149–155

    2008

  15. [15]

    Adrien Bibal, Rémi Cardon, David Alfter, Rodrigo Wilkens, Xiaoou Wang, Thomas François, and Patrick Watrin. 2022. Is attention explanation? an introduction to the debate. InProceedings of the 60th Annual Meeting of the Association for Computational Linguistics (volume 1: long papers). 3889–3900

    2022

  16. [16]

    Tobias Braun, Mark Rothermel, Marcus Rohrbach, and Anna Rohrbach. 2025. DEFAME: Dynamic Evidence-based FAct-checking with Multimodal Experts. In International Conference on Machine Learning. PMLR, 5383–5417

    2025

  17. [17]

    Alexander S Choi, Syeda Sabrina Akter, JP Singh, and Antonios Anastasopoulos

  18. [18]

    InProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing

    The LLM effect: Are humans truly using LLMs, or are they being influenced by them instead?. InProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 22032–22054

    2024

  19. [19]

    Yuwei Chuai, Moritz Pilarski, Gabriele Lenzini, and Nicolas Pröllochs. 2024. Community notes reduce the spread of misleading posts on X. (2024)

    2024

  20. [20]

    Yuwei Chuai, Haoye Tian, Nicolas Pröllochs, and Gabriele Lenzini. 2024. Did the roll-out of community notes reduce engagement with misinformation on X/Twitter?Proceedings of the ACM on human-computer interaction8, CSCW2 (2024), 1–52

    2024

  21. [21]

    Yuwei Chuai, Shuning Zhang, Ziming Wang, Xin Yi, Mohsen Mosleh, and Gabriele Lenzini. 2025. Request a Note: How the Request Function Shapes X’s Community Notes System.arXiv preprint arXiv:2509.09956(2025)

    arXiv 2025

  22. [22]

    Sunhao Dai, Yuqi Zhou, Liang Pang, Weihao Liu, Xiaolin Hu, Yong Liu, Xiao Zhang, Gang Wang, and Jun Xu. 2024. Neural retrievers are biased towards llm-generated content. InProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 526–537

    2024

  23. [23]

    Laurence Dierickx and Carl-Gustav Lindén. 2023. Journalism and fact-checking technologies: Understanding user needs.communication+ 110, 1 (2023)

    2023

  24. [24]

    Tim Draws, David La Barbera, Michael Soprano, Kevin Roitero, Davide Ceolin, Alessandro Checco, and Stefano Mizzaro. 2022. The effects of crowd worker biases in fact-checking tasks. InProceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency. 2114–2124

    2022

  25. [25]

    Fiona Draxler, Anna Werner, Florian Lehmann, Matthias Hoppe, Albrecht Schmidt, Daniel Buschek, and Robin Welsch. 2024. The AI ghostwriter effect: When users do not perceive ownership of AI-generated text but self-declare as authors.ACM Transactions on Computer-Human Interaction31, 2 (2024), 1–40

    2024

  26. [26]

    Ziv Epstein, Nicolo Foppiani, Sophie Hilgard, Sanjana Sharma, Elena Glassman, and David Rand. 2022. Do explanations increase the effectiveness of AI-crowd generated fake news warnings?. InProceedings of the International AAAI Confer- ence on Web and Social Media, Vol. 16. 183–193

    2022

  27. [27]

    Jonathan L Freedman and David O Sears. 1965. Selective exposure. InAdvances in experimental social psychology. Vol. 2. Elsevier, 57–97

    1965

  28. [28]

    Liye Fu, Benjamin Newman, Maurice Jakesch, and Sarah Kreps. 2023. Com- paring sentence-level suggestions to message-level suggestions in AI-mediated communication. InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems. 1–13

    2023

  29. [29]

    The Data Says Otherwise

    Yu Fu, Shunan Guo, Jane Hoffswell, Victor S. Bursztyn, Ryan Rossi, and John Stasko. 2024. " The Data Says Otherwise"—Towards Automated Fact-checking and Communication of Data Claims. InProceedings of the 37th Annual ACM Symposium on User Interface Software and Technology. 1–20

    2024

  30. [30]

    Daniel Fürst, Mennatallah El-Assady, Daniel A Keim, and Maximilian T Fis- cher. 2025. Challenges and Opportunities for Visual Analytics in Jurisprudence. Artificial Intelligence and Law(2025), 1–32

    2025

  31. [31]

    William Godel, Zeve Sanderson, Kevin Aslett, Jonathan Nagler, Richard Bonneau, Nathaniel Persily, and Joshua A Tucker. 2021. Moderating with the mob: Eval- uating the efficacy of real-time crowdsourced fact-checking.Journal of Online Trust and Safety1, 1 (2021)

    2021

  32. [32]

    Lucas Graves and Michelle Amazeen. 2019. Fact-checking as idea and practice in journalism. (2019)

    2019

  33. [33]

    2025.Towards Automating Citation Worthiness Detection for Sen- tences in Low-Resource Languages

    Aida Halitaj. 2025.Towards Automating Citation Worthiness Detection for Sen- tences in Low-Resource Languages. Ph. D. Dissertation. Queen Mary University of Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Trovato et al. London

    2025

  34. [34]

    Sandra G Hart and Lowell E Staveland. 1988. Development of NASA-TLX (Task Load Index): Results of empirical and theoretical research. InAdvances in psy- chology. Vol. 52. Elsevier, 139–183

    1988

  35. [35]

    Naeemul Hassan, Fatma Arslan, Chengkai Li, and Mark Tremayne. 2017. Toward automated fact-checking: Detecting check-worthy factual claims by claimbuster. InProceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining. 1803–1812

    2017

  36. [36]

    Gaole He, Stefan Buijsman, and Ujwal Gadiraju. 2023. How stated accuracy of an AI system and analogies to explain accuracy affect human reliance on the system. Proceedings of the ACM on Human-Computer Interaction7, CSCW2 (2023), 1–29

    2023

  37. [37]

    Gaole He, Lucie Kuiper, and Ujwal Gadiraju. 2023. Knowing about knowing: An illusion of human competence can hinder appropriate reliance on AI systems. InProceedings of the 2023 CHI conference on human factors in computing systems. 1–18

    2023

  38. [38]

    Andreas Hinderks. 2017. Design and evaluation of a short version of the user experience questionnaire (UEQ-S).International Journal of Interactive Multimedia and Artificial Intelligence(2017)

    2017

  39. [39]

    Qisheng Hu, Quanyu Long, and Wenya Wang. 2025. Decomposition Dilemmas: Does Claim Decomposition Boost or Burden Fact-Checking Performance?. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 6313–6336

    2025

  40. [40]

    Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, et al. 2025. A survey on hallucination in large language models: Principles, taxonomy, chal- lenges, and open questions.ACM Transactions on Information Systems43, 2 (2025), 1–55

    2025

  41. [41]

    Alon Jacovi and Yoav Goldberg. 2020. Towards faithfully interpretable NLP systems: How should we define and evaluate faithfulness?. InProceedings of the 58th annual meeting of the association for computational linguistics. 4198–4205

    2020

  42. [42]

    Farnaz Jahanbakhsh, Yannis Katsis, Dakuo Wang, Lucian Popa, and Michael Muller. 2023. Exploring the use of personalized AI for identifying misinformation on social media. InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems. 1–27

    2023

  43. [43]

    Eunchae Jang, Hui Min Lee, Sangwook Lee, Yongnam Jung, and S Shyam Sundar

  44. [44]

    InProceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems

    Too good to be false: How photorealism promotes susceptibility to mis- information. InProceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–8

  45. [45]

    Jiun-Yin Jian, Ann M Bisantz, and Colin G Drury. 2000. Foundations for an empirically determined scale of trust in automated systems.International journal of cognitive ergonomics4, 1 (2000), 53–71

    2000

  46. [46]

    Junqi Jiang, Tom Bewley, Salim I Amoukou, Francesco Leofante, Antonio Rago, Saumitra Mishra, and Francesca Toni. 2025. Representation Consistency for Accurate and Coherent LLM Answer Aggregation. InThe Thirty-ninth Annual Conference on Neural Information Processing Systems

    2025

  47. [47]

    Yifan Jiang, Cong Zhang, Bofei Zhang, Yifan Yang, Bingzhang Wang, and Yew- Soon Ong. 2026. From Pixels to Facts (Pix2Fact): Benchmarking Multi-Hop Reasoning for Fine-Grained Visual Fact Checking.arXiv preprint arXiv:2602.00593 (2026)

    Pith/arXiv arXiv 2026

  48. [48]

    Prerna Juneja and Tanushree Mitra. 2022. Human and technological infrastruc- tures of fact-checking.Proceedings of the ACM on Human-Computer Interaction6, CSCW2 (2022), 1–36

    2022

  49. [49]

    Jiho Kim, Sungjin Park, Yeonsu Kwon, Yohan Jo, James Thorne, and Edward Choi

  50. [50]

    InProceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

    FactKG: Fact verification via reasoning on knowledge graphs. InProceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 16190–16206

  51. [51]

    Mio Kobayashi, Ai Ishii, Chikara Hoshino, Hiroshi Miyashita, and Takuya Mat- suzaki. 2017. Automated historical fact-checking by passage retrieval, word statistics, and virtual question-answering. InProceedings of the Eighth Interna- tional Joint Conference on Natural Language Processing (Volume 1: Long Papers). 967–975

    2017

  52. [52]

    Allison Koenecke, Jed Stiglitz, David Mimno, and Matthew Wilkens. 2025. Tasks and roles in legal ai: Data curation, annotation, and verification.arXiv preprint arXiv:2504.01349(2025)

    arXiv 2025

  53. [53]

    Neema Kotonya and Francesca Toni. 2020. Explainable automated fact-checking for public health claims. InProceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). 7740–7754

    2020

  54. [54]

    David La Barbera, Kevin Roitero, Stefano Mizzaro, et al. 2022. A Hybrid Human- In-The-Loop Framework for Fact Checking.. InNL4AI@ AI* IA. 13–23

    2022

  55. [55]

    Duo Lan, Yicheng Zhu, Meiyu Liu, and Chuge He. 2025. AI Agency in Fact- Checking: Role-Based Machine Heuristics and Publics’ Conspiratorial Orienta- tion.Media and Communication13 (2025)

    2025

  56. [56]

    Hao-Ping Lee, Advait Sarkar, Lev Tankelevitch, Ian Drosos, Sean Rintel, Richard Banks, and Nicholas Wilson. 2025. The impact of generative AI on critical thinking: Self-reported reductions in cognitive effort and confidence effects from a survey of knowledge workers. InProceedings of the 2025 CHI conference on human factors in computing systems. 1–22

    2025

  57. [57]

    Haiwen Li, Soham De, Manon Revel, Andreas Haupt, Brad Miller, Keith Coleman, Jay Baxter, Martin Saveski, and Michiel Bakker. 2025. Scaling Human Judgment in Community Notes with LLMs.Journal of Online Trust and Safety3, 1 (2025)

    2025

  58. [58]

    Q Vera Liao and Kush R Varshney. 2021. Human-centered explainable ai (xai): From algorithms to user experiences.arXiv preprint arXiv:2110.10790(2021)

    arXiv 2021

  59. [59]

    Gionnieve Lim and Simon T Perrault. 2023. XAI in Automated Fact-Checking? The Benefits Are Modest and There’s No One-Explanation-Fits-All. InProceedings of the 35th Australian Computer-Human Interaction Conference. 624–638

    2023

  60. [60]

    Houjiang Liu, Anubrata Das, Alexander Boltz, Didi Zhou, Daisy Pinaroc, Matthew Lease, and Min Kyung Lee. 2024. Human-centered NLP Fact-checking: Co- Designing with Fact-checkers using Matchmaking for AI.Proceedings of the ACM on Human-Computer Interaction8, CSCW2 (2024), 1–44

    2024

  61. [61]

    Houjiang Liu, Jacek Gwizdka, and Matthew Lease. 2025. Exploring Multidimen- sional Checkworthiness: Designing AI-assisted Claim Prioritization for Human Fact-checkers.Proceedings of the ACM on Human-Computer Interaction9, 7 (2025), 1–49

    2025

  62. [62]

    Varun Magesh, Faiz Surani, Matthew Dahl, Mirac Suzgun, Christopher D Man- ning, and Daniel E Ho. 2025. Hallucination-free? Assessing the reliability of leading AI legal research tools.Journal of empirical legal studies22, 2 (2025), 216–242

    2025

  63. [63]

    David M Markowitz, Timothy R Levine, Kim B Serota, and Alivia D Moore. 2023. Cross-checking journalistic fact-checkers: The role of sampling and scaling in interpreting false and misleading statements.Plos one18, 7 (2023), e0289004

    2023

  64. [64]

    Nora McDonald, Sarita Schoenebeck, and Andrea Forte. 2019. Reliability and inter-rater reliability in qualitative research: Norms and guidelines for CSCW and HCI practice.Proceedings of the ACM on human-computer interaction3, CSCW (2019), 1–23

    2019

  65. [65]

    Nicholas Micallef, Vivienne Armacost, Nasir Memon, and Sameer Patil. 2022. True or false: Studying the work practices of professional fact-checkers.Proceedings of the ACM on Human-Computer Interaction6, CSCW1 (2022), 1–44

    2022

  66. [66]

    Serge Moscovici and Claude Faucheux. 1972. Social influence, conformity bias, and the study of active minorities. InAdvances in experimental social psychology. Vol. 6. Elsevier, 149–202

    1972

  67. [67]

    2025.Multi-Stage LLM Reasoning for Automated Detection and Classification of High-Impact Misinformation

    Anushka Manchanda Nair. 2025.Multi-Stage LLM Reasoning for Automated Detection and Classification of High-Impact Misinformation. Ph. D. Dissertation. Massachusetts Institute of Technology

    2025

  68. [68]

    P Nakov, D Corney, M Hasanain, F Alam, T Elsayed, A Barron-Cedeno, P Papotti, S Shaar, G Da San Martino, et al. 2021. Automated Fact-Checking for Assisting Human Fact-Checkers. InIJCAI. International Joint Conferences on Artificial Intelligence, 4551–4558

    2021

  69. [69]

    An T Nguyen, Aditya Kharosekar, Saumyaa Krishnan, Siddhesh Krishnan, Eliza- beth Tate, Byron C Wallace, and Matthew Lease. 2018. Believe it or not: designing a human-ai partnership for mixed-initiative fact-checking. InProceedings of the 31st annual ACM symposium on user interface software and technology. 189–199

    2018

  70. [70]

    Margit E Oswald and Stefan Grosjean. 2004. Confirmation bias.Cognitive illusions: A handbook on fallacies and biases in thinking, judgement and memory79 (2004), 83

    2004

  71. [71]

    Liangming Pan, Xiaobao Wu, Xinyuan Lu, Luu Anh Tuan, William Yang Wang, Min-Yen Kan, and Preslav Nakov. 2023. Fact-checking complex claims with program-guided reasoning. InProceedings of the 61st annual meeting of the asso- ciation for computational linguistics (volume 1: Long papers). 6981–7004

    2023

  72. [72]

    Is this professionally correct?

    Brandon Patterson, Anne R Diekema, Elizabeth Betsy S Hopkins, Duane Wilson, and Nena Schvaneveldt. 2025. “Is this professionally correct?”: understanding the criteria nurses use to evaluate information.Journal of the Medical Library Association: JMLA113, 4 (2025), 298

    2025

  73. [73]

    Vitali Petsiuk, Rajiv Jain, Varun Manjunatha, Vlad I Morariu, Ashutosh Mehra, Vicente Ordonez, and Kate Saenko. 2021. Black-box explanation of object detec- tors via saliency maps. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 11443–11452

    2021

  74. [74]

    Kashyap Popat, Subhabrata Mukherjee, Andrew Yates, and Gerhard Weikum

  75. [75]

    InProceedings of the 2018 conference on empirical methods in natural language processing

    Declare: Debunking fake news and false claims using evidence-aware deep learning. InProceedings of the 2018 conference on empirical methods in natural language processing. 22–32

    2018

  76. [76]

    Subhey Sadi Rahman, Md Adnanul Islam, Md Mahbub Alam, Musarrat Zeba, Md Abdur Rahman, Sadia Sultana Chowa, Mohaimenul Azam Khan Raiaan, and Sami Azam. 2026. Hallucination to truth: a review of fact-checking and factuality evaluation in large language models.Artificial Intelligence Review(2026)

    2026

  77. [77]

    Why should i trust you?

    Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin. 2016. " Why should i trust you?" Explaining the predictions of any classifier. InProceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining. 1135–1144

    2016

  78. [78]

    Giuseppe Romeo and Daniela Conti. 2026. Exploring automation bias in human– AI collaboration: a review and implications for explainable AI.AI & SOCIETY41, 1 (2026), 259–278

    2026

  79. [79]

    Mohammed Saeed, Nicolas Traub, Maelle Nicolas, Gianluca Demartini, and Paolo Papotti. 2022. Crowdsourced fact-checking at Twitter: How does the crowd compare with experts?. InProceedings of the 31st ACM international conference on information & knowledge management. 1736–1746. CANote: Empowering Fact-checking Note Writing Through Scaffolded and Provenance...

    2022

  80. [80]

    Max Schemmer, Niklas Kühl, Carina Benz, and Gerhard Satzger. 2022. On the In- fluence of Explainable AI on Automation Bias. InProceedings of the 30th European Conference on Information Systems (ECIS), Timis,oara, RO, June 18-24, 2022

    2022

Showing first 80 references.