Pith. sign in

REVIEW 3 major objections 5 minor 68 references

Fact-Check Your Information (FYI): A Design Probe to Understand How People Actually Fact-Check Data-Driven Articles

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Readers do not follow a fixed fact-checking pipeline: in a 22-person study they flexibly composed AI-first, manual-first, and parallel workflows, used self-built charts to audit AI conclusions, and calibrated trust by cross-tool agreement.

desk verdict Useful design-probe study with open artifacts and honest limitations; the central claims are plausible but the task design partly pre-arranges what it observes. read the letter →

arxiv 2608.06804 v1 pith:V2I34YRD submitted 2026-08-07 cs.HC

Nguyen-Truong Thinh , Yuxuan Du , Phongsakon Mark Konrad , Arpit Narechania This is my paper · ORCID
classification cs.HC
keywords dataclaimsfact-checkinghuman-AIinteractiontrustcalibrationvisualizationdesignprobelargelanguagemodelsmixed-initiativesystems
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

FYI is a browser extension that lets readers fact-check statistical statements in online articles against the underlying dataset, using four tools that range from full automation (an Auto Check pipeline and an AI chat) to hands-on inspection (a sortable table and a self-service chart builder). The paper uses FYI as an instrumented probe to answer a question prior systems have left open: when both AI and manual verification are available at once, how do people actually detect, verify, and judge data claims? In a 22-participant exploratory study, readers did not march through a fixed detection-verification-determination pipeline. Instead they composed three recurring workflow archetypes—AI-first with manual confirmation, manual-first with AI supplement, and parallel co-review—and they treated self-built visualizations as the primary mechanism for checking AI conclusions. The paper argues that trust in AI is calibrated dynamically, growing when independent tools converge and eroding when outputs disagree, and that future fact-checking systems should therefore treat AI as a starting point rather than an authority, while elevating visualization to a core verification capability.

What carries the argument

FYI is a browser extension that embeds the whole fact-checking workspace in a side panel beside the article under review. Its load-bearing design is a spectrum of four verification tools sharing one uploaded dataset: Auto Check (a four-step automated pipeline that streams an evidence chart and verdict), AI Chat (a multi-turn LLM with optional web search and client-side Python data analysis), Table Explorer (direct sorting and filtering of raw rows), and Chart Builder (a shelf-based chart authoring tool). A 25-type interaction logger records every user action as a timestamped event, so the reading session itself becomes observable data. This combination—multiple complementary modalities plus full provenance logging—is what allows the authors to characterize workflow archetypes and trust calibration rather than measuring only end-task accuracy.

What would settle it

Run the same FYI study with a sample matched to the target population (lower self-rated visualization literacy and sporadic generative-AI use) on articles that require multi-table joins: if most participants follow one fixed tool order, rarely open Chart Builder after Auto Check, or accept AI verdicts without inspecting the data, the three-archetype and visual-auditing claims fail to generalize.

Watch

Extended reading notes

Core claim

The paper's central empirical claim is that mixed-initiative data fact-checking is neither automation-led nor manual-led but a fluid, multi-modal activity. Analyzing 2,250 logged interactions, think-aloud protocols, and interviews from 22 participants, the authors identify three workflow archetypes—AI-first with manual confirmation (9/22), manual-first with AI supplement (6/22), and parallel co-review (4/22)—and show that individuals often moved between archetypes within a session. Chart Builder was the most-used per-claim tool (105 of 156 investigated claims), and in 55 instances participants deliberately opened it after Auto Check to audit the AI's conclusion; a self-built chart that aligned with or refuted a claim often served as the stopping criterion for a verdict, sometimes overriding AI outputs. Trust in AI shifted with experience: convergence of independent tools raised confidence, numerical inconsistencies eroded it, and one early AI error could collapse trust for the whole session. The authors also document a dominant 'AI initiates, human decides' reliance model, with Auto Check broadly accurate on straightforward claims (71% agreement with ground truth) but blind to the one claim requiring contextual judgment that the dataset could not settle, which participants were better at flagging as unverifiable.

Load-bearing premise

The load-bearing premise is that the 22 study participants—university-affiliated, highly visualization-literate, and daily generative-AI users—behave like the lay readers with basic data and visualization literacy that FYI is designed for, when both groups fact-check a single data-driven article.

Editorial extensions

If this is right

  • Fact-checking systems should present AI verdicts as provisional hypotheses that invite human confirmation, rather than as definitive answers that require effort to override.
  • Visualization authoring should be treated as a primary auditing capability, with AI-assisted chart suggestions offered to users who cannot easily build charts themselves.
  • Interfaces should offer independent, freely composable tools instead of rigid step-by-step wizards, because participants adapted their workflow across claims and even within a single session.
  • Systems should expose intermediate reasoning, data queries, generated code, and confidence scores, since visible process supported trust calibration and selective override.
  • Designers of trust should expect calibration to be brittle: one inconsistent numeric output can outweigh many successes, so cross-tool consistency matters as much as average accuracy.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If these workflow archetypes generalize, accuracy reporting for fact-checking tools should include tool-sequence analyses, not just verdict correctness, because a tool's value depends on how readers interleave it with other modalities.
  • The usability-confidence paradox suggests a concrete design test: giving low-literacy users AI-assisted chart suggestions should increase their willingness to audit AI outputs, and a controlled comparison of verdict confidence and override behavior could verify this.
  • The observed AI-on-AI cross-checks imply that future systems should distinguish 'grounded in raw data' from 'two language models happen to agree,' since agreement between LLMs is weaker evidence than a user-built chart derived from the dataset.
  • A natural extension is to run the same probe on finance or public-health articles whose claims require joins across tables; the plausible prediction is more manual-first behavior and more unverifiable verdicts, but the paper itself does not establish that.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper presents FYI, an open-source browser extension that embeds claim detection, four verification tools (Auto Check, AI Chat, Table Explorer, Chart Builder), and user-authored verdicts into the reading environment. Using FYI as an instrumented design probe, the authors ran an exploratory N=22 study in which participants fact-checked six embedded data claims in a movie-data article while thinking aloud. Based on interaction logs, transcripts, and questionnaires, they report three workflow archetypes (AI-first with manual confirmation; manual-first with AI supplement; parallel co-review), dynamic trust calibration driven by cross-tool convergence and inconsistency, and the use of self-built charts as a primary mechanism for auditing AI outputs. They derive four design implications for mixed-initiative fact-checking systems and release the prototype and study materials.

Significance. If the behavioral findings hold, the paper makes a useful empirical contribution to human-AI sensemaking and fact-checking research: it moves beyond system-capability papers to document real tool-composition behavior, and it provides a reusable open-source testbed with fine-grained interaction logging, prompts, and raw logs. The triangulation of interaction logs, think-aloud protocols, and interviews is a strength, as is the transparent reporting of verdict accuracy against researcher ground truth in Sec. 5.5. The main significance is as an exploratory design probe; the contribution is descriptive rather than confirmatory, and the implications (DI1–DI4) are tied to the observed behaviors in a way that should generalize only if the identified threats are addressed.

major comments (3)
  1. [Sec. 4.2] Section 4.2 states that the six embedded claims were designed 'to require participants to combine multiple verification strategies rather than rely on any single tool or pathway.' Because the stimulus was deliberately constructed so that no single tool suffices, the observation that participants composed multi-tool workflows (Sec. 5.3) and that three archetypes emerged is in part an artifact of the experimental instrument, not an independent fact about reader behavior. To support the central 'actually fact-check' claim, the authors should either include claims answerable by a single tool or direct table lookup, conduct a per-claim analysis showing that the archetypes and chart-auditing behavior also hold for simple claims, or explicitly re-scope the contribution to behavior under a task that demands multi-tool composition. This is load-bearing because the paper's main empirical result—flexible multi-tool composition—depends on it.
  2. [Sec. 4.1 and Sec. 7] The stated target population is 'laypersons with basic data and visualization literacy' (Sec. 1 and Sec. 4.1), but the sample reports high visualization literacy (M=5.68/7) and near-daily generative AI use (M=6.18/7). The conclusions that 'visualization serves as the primary auditing mechanism' (Sec. 5.4, T7) and that limited visualization literacy creates a 'usability–confidence paradox' (Sec. 6.3) are therefore likely gated by the sample's visualization skill; the paper itself notes that P17 avoided Chart Builder (Sec. 5.4) and acknowledges generalizability limits (Sec. 7). To make the population claim, the authors should either recruit a broader sample, stratify the analyses by data literacy, or reframe the contribution as an account of data-literate users' behavior with implications, rather than evidence, for laypersons.
  3. [Sec. 5.3 and Table 4] The three workflow archetypes (T3–T5) are central to the paper's contribution, but the method for assigning participants to archetypes is not reported. Table 4 lists participant IDs per theme, yet the paper does not state whether classification was done per participant across the whole session or per claim, what criteria or thresholds define each archetype, or whether the three coders agreed on these assignments. Figure 2 shows per-claim sequences while the text reports archetype counts (9/6/4) as session-level categories, making the taxonomy hard to audit. Please provide the operational coding scheme, a per-participant classification rule, and inter-rater agreement statistics, or present the archetypes as per-claim orientations rather than participant types.
minor comments (5)
  1. [Sec. 2.1 and Sec. 2.2] There are unresolved citation placeholders: '[?]' appears in Sec. 2.1 for the statement that data claims implicitly refer to an underlying dataset, and '[?]' appears in Sec. 2.2 for the Thucy system, which corresponds to reference [53]. These should be replaced with proper citations.
  2. [Figure 2] The caption says the figure covers 131 claims, while Sec. 5.1 reports 156 actively investigated claims; please clarify whether the figure excludes claims without tool-use sequences and state the inclusion criterion.
  3. [Sec. 4.4] The thematic analysis section describes independent coding by three researchers but reports no inter-rater reliability measure; for a qualitative exploratory study this is acceptable, but a brief statement on consensus and stability of the final themes would strengthen the reporting.
  4. [Sec. 5.5] The accuracy comparison (n=79) includes only verdict instances for which researcher ground truth, Auto Check, and participant labels were all available; please clarify how the excluded instances were distributed and whether this selection affects the reported agreement rates.
  5. [Sec. 3.1] The design goals are clear, but DG2 states that 'web search provides external corroboration' while web search is implemented only as a toggle within AI Chat; the paper could note this placement earlier to avoid implying a fifth standalone tool.

Circularity Check

1 steps flagged · score 3.0 of 10

Partial design-induced circularity: multi-tool workflow observation is guaranteed by the stimulus design, but archetypes and trust dynamics rest on independent logs.

  1. other [Sec. 4.2 (Materials) and Sec. 5.3 (RQ2)]
    "We further designed the claims so that no single tool would suffice to check them: some can be settled by a direct data operation such as filtering or charting a value, others require multi-step analysis such as comparing groups or computing a correlation, and others demand judgment beyond the data... This variation was intended to require participants to combine multiple verification strategies rather than rely on any single tool or pathway."

    The instrument was explicitly engineered so that no single tool could verify any embedded claim. Reporting 'All four verification tools achieved broad adoption... participants used 3.7 out of 4 tools' and describing AI-first/manual-confirmation and manual-first/AI-supplement workflows as findings partly restates this design constraint: a workflow using a single tool was not a viable option for the given claims. The specific ordering archetypes, parallel co-review, and trust-calibration dynamics are not entailed by the design, so the circularity is partial rather than total.

full rationale

The paper is an exploratory design-probe study, not a derivation; most of its claims are empirical observations. The three workflow archetypes (T3-T5), visual-auditing pattern, and dynamic trust calibration are supported by independent interaction logs, tool-transition graphs, and participant quotes, so they do not reduce to fitted parameters or to self-citation. Self-citations (WebSeek, NL4DV, ProvenanceWidgets) are contextual and not load-bearing; no uniqueness theorem or ansatz is imported from prior work. The main circularity concern is the deliberate task construction in Sec. 4.2: claims were designed to require combining multiple verification strategies, so observing multi-tool composition is partly an artifact of the stimulus rather than a free behavioral discovery. The paper's own limitations section acknowledges sample and domain constraints but does not fully neutralize this design-level guarantee. This warrants a moderate partial score of 3, not higher, because the central qualitative content (which tools lead, how trust shifts, and how charts audit AI) remains grounded in logged sequences and think-aloud evidence.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claims rest on qualitative coding and a designed probe rather than a formal derivation. No free parameters are fitted. Four domain assumptions are load-bearing: the ground-truth labels, the truthfulness of self-reports, dataset integrity, and model output stability.

assumptions (4)
  • domain assumption Researcher-assigned ground-truth labels for the six embedded data claims are correct and unambiguous.
    Used in Sec. 5.5 to compute Auto Check and participant agreement rates; if labels are contestable (e.g., the 'unverifiable' claim), the accuracy comparison weakens.
  • domain assumption Participants' think-aloud comments and post-study interviews truthfully reflect their reasoning and trust states.
    Thematic analysis in Sec. 5.2-5.5 relies on these self-reports; think-aloud can miss tacit processes and interviews can be subject to social desirability.
  • domain assumption The constructed movie dataset is accurate and sufficient to verify the embedded claims.
    Sec. 4.2 states the dataset was 'constructed from publicly available sources'; no independent verification of dataset integrity is provided.
  • domain assumption GPT-4.1's outputs in Auto Check and AI Chat are stable enough to support the reported error patterns.
    Sec. 7 acknowledges outputs vary with model version and stochastic sampling; the study's accuracy findings are tied to one model instance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Fact-Check Your Information (FYI): A Design Probe to Understand How People Actually Fact-Check Data-Driven Articles." pith.science (2026). https://pith.science/paper/V2I34YRD

@misc{pith2026260806804,
  author       = {Pith},
  title        = {Pith review of: Fact-Check Your Information (FYI): A Design Probe to Understand How People Actually Fact-Check Data-Driven Articles},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/V2I34YRD}},
  note         = {Machine review of arXiv:2608.06804}
}
read the original abstract

Data-driven journalism and policy reports frequently rely on statements grounded in statistical evidence, referred to as data claims. Verifying such a claim requires connecting it to the underlying structured dataset. However, existing systems typically isolate automated fact-checking from manual data exploration, leaving it unclear how readers coordinate AI assistance with manual inspection of the evidence in practice. We present FYI, a browser extension that embeds fact-checking in the reading environment, and use it as a design probe to study how people detect, verify, and determine the validity of data claims against the underlying dataset. FYI provides four complementary tools spanning the spectrum from full automation to manual data exploration. In an exploratory study (N=22), participants used FYI to fact-check claims in a data-driven article. We find that participants adopted three distinct workflow archetypes---AI-first with manual confirmation, manual-first with AI supplement, and parallel co-review---with visualization serving as the primary mechanism for auditing AI conclusions. Trust in AI shifted dynamically, growing when multiple tools converged and eroding when AI outputs were inconsistent. These findings suggest that fact-checking systems should treat AI as a starting point that human verification complements rather than a definitive authority, elevate visualization as a core verification capability, and support flexible, user-driven workflows. We release FYI as open-source software for further research at https://github.com/DataVisards/FYI.

Figures

Figures reproduced from arXiv: 2608.06804 by the authors.

Figure 1
Figure 1. The interface of FYI, a browser extension that enables in-situ data claim fact-checking directly alongside the article under [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 3
Figure 3. Transitions between tools (289 transitions across 131 claims). [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Verdict agreement with researcher ground truth ( [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

68 extracted references · 26 canonical work pages

  1. [1]

    Aljamaan, M.-H

    F. Aljamaan, M.-H. Temsah, I. Altamimi, A. Al-Eyadhy, A. Jamal, K. Al- hasan et al. Reference hallucination score for medical artificial intelligence chatbots: Development and usability study.JMIR Med. Inform., 12, art. no. e54345, Jul. 2024. doi: 10.2196/54345 2

    doi:10.2196/54345 2024

  2. [2]

    P. M. Allen, J. A. Edwards, F. J. Snyder, K. A. Makinson, and D. M. Hamby. The effect of cognitive load on decision making with graphically displayed uncertainty information.Risk Anal., 34(8):1495–1505, Aug

    work page

  3. [3]

    Aslett, Z

    K. Aslett, Z. Sanderson, W. Godel, N. Persily, J. Nagler, and J. A. Tucker. Online searches to evaluate misinformation can increase its perceived veracity.Nature, 625:548–556, Jan. 2024. doi: 10.1038/s41586-023 -06883-y 4

    doi:10.1038/s41586-023 2024

  4. [4]

    Balalau, S

    O. Balalau, S. Ebel, T. Galizzi, I. Manolescu, Q. Massonnat, A. Deiana et al. Statistical claim checking: StatCheck in action. InProc. ACM Int. Conf. Inf. & Knowl. Manage. (CIKM), pp. 4798–4802. ACM, New York, Oct. 2022. doi: 10.1145/3511808.3557198 3

    arXiv 2022

  5. [5]

    J. E. Block, S. Esmaeili, E. D. Ragan, J. R. Goodall, and G. D. Richard- son. The influence of visual provenance representations on strategies in a collaborative hand-off data analysis scenario.IEEE Trans. Visual. Com- put. Graphics, 29(1):1113–1123, Jan. 2023. doi: 10.1109/TVCG.2022. 3209495 4

    doi:10.1109/tvcg.2022 2023

  6. [6]

    Botnevik, E

    B. Botnevik, E. Sakariassen, and V . Setty. BRENDA: Browser extension for fake news detection. InProc. ACM Int. Conf. Res. Develop. Inf. Retrieval (SIGIR), SIGIR ’20, pp. 2117–2120. ACM, New York, Jul. 2020. doi: 10.1145/3397271.3401396 2

    arXiv 2020

  7. [7]

    Braun and V

    V . Braun and V . Clarke. Using thematic analysis in psychology.Qual. Res. Psychol., 3(2):77–101, Jan. 2006. doi: 10.1191/1478088706qp063oa 6

    doi:10.1191/1478088706qp063oa 2006

  8. [8]

    Cao and C.-M

    S. Cao and C.-M. Huang. Understanding user reliance on AI in assisted decision-making.Proc. ACM Hum.-Comput. Interact., 6(CSCW2), art. no. 471, 23 pp., Nov. 2022. doi: 10.1145/3555572 3, 8

    doi:10.1145/3555572 2022

Show all 68 references
  1. [9]

    B. G. d. S. Cezar and A. C. G. Maçada. Cognitive overload, anxiety, cogni- tive fatigue, avoidance behavior and data literacy in big data environments. Inf. Process. & Manage., 60(6), art. no. 103482, Nov. 2023. doi: 10.1016/j .ipm.2023.103482 2

    2023

  2. [10]

    J. H. Chae and D. Tewksbury. Perceiving AI intervention does not compro- mise the persuasive effect of fact-checking.New Media Soc., 28(1):191– 211, Jan. 2026. doi: 10.1177/14614448241286881 3

    2026 doi

  3. [11]

    Chegini, K

    A. Chegini, K. Rezaei, H. Eghbalzadeh, and S. Feizi. RePanda: Pandas- powered tabular verification and reasoning. InProc. 63rd Annu. Meeting Assoc. Comput. Linguistics (ACL), pp. 32200–32212. Association for Computational Linguistics, Vienna, Austria, Jul. 2025. doi: 10.18653/...

    2025 doi

  4. [12]

    W. Chen, H. Wang, J. Chen, Y . Zhang, H. Wang, S. Li et al. TabFact: A large-scale dataset for table-based fact verification. InProc. Int. Conf. Learn. Representations (ICLR). Virtual Conference, Jun. 2020. doi: 10. 48550/arXiv.1909.02164 2, 3

    2020 doi arXiv

  5. [13]

    Chen and H

    Z.-T. Chen and H. Xia. CrossData: Leveraging text-data connections for authoring data documents. InProc. ACM CHI Conf. Human Factors Comput. Syst. (CHI), CHI ’22, art. no. 95. ACM, New York, Apr. 2022. doi: 10.1145/3491102.3517485 3

    2022

  6. [14]

    Cheng, T

    Z. Cheng, T. Xie, P. Shi, C. Li, R. Nadkarni, Y . Hu et al. Binding language models in symbolic languages. InProc. Int. Conf. Learn. Representations (ICLR). Kigali, Rwanda, May 2023. doi: 10.48550/arXiv.2210.02875 2, 3, 5

    2023 doi arXiv

  7. [15]

    A. K. Das and K. Mueller. MisVisFix: An interactive dashboard for detecting, explaining, and correcting misleading visualizations using large language models.IEEE Trans. Visual. Comput. Graphics, 32(1):134–144, Jan. 2026. doi: 10.1109/TVCG.2025.3633884 3, 4

    2026

  8. [16]

    M. R. DeVerna, H. Y . Yan, K.-C. Yang, and F. Menczer. Fact-checking information from large language models can decrease headline discern- ment.Proc. Natl. Acad. Sci. USA, 121(50), art. no. e2322823121, Dec

  9. [17]

    K. A. Ericsson and H. A. Simon.Protocol Analysis: Verbal Reports as Data. MIT Press, Cambridge, MA, USA, rev. ed., 1993. doi: 10. 7551/mitpress/5657.001.0001 6

    1993

  10. [18]

    The data says otherwise

    Y . Fu, S. Guo, J. Hoffswell, V . S. Bursztyn, R. Rossi, and J. Stasko. "The data says otherwise" — towards automated fact-checking and communica- tion of data claims. InProc. ACM Symp. User Interface Softw. Technol. (UIST), UIST ’24, art. no. 134, 20 pp. ACM, New York, Oct. 2...

    2024

  11. [19]

    Goddard, A

    K. Goddard, A. Roudsari, and J. C. Wyatt. Automation bias: a systematic review of frequency, effect mediators, and mitigators.J. Amer. Med. Inform. Assoc., 19(1):121–127, Jan. 2012. doi: 10.1136/amiajnl-2011-000089 2, 3

    2012 doi

  12. [20]

    L. Graves. Understanding the promise and limits of automated fact- checking. Factsheet, Reuters Inst. Study Journalism, Univ. Oxford, Oxford, U.K., Feb. 2018. doi: 10.60625/risj-nqnx-bg89 2

    2018 doi

  13. [21]

    Z. Gu, J. Fan, N. Tang, P. Nakov, X. Zhao, and X. Du. PASTA: Table- operations aware fact verification via sentence-table cloze pre-training. InProc. Conf. Empirical Methods Natural Lang. Process. (EMNLP), pp. 4971–4983. Association for Computational Linguistics, Abu Dhabi, Un...

    2022 doi

  14. [22]

    Z. Guo, M. Schlichtkrull, and A. Vlachos. A survey on automated fact- checking.Trans. Assoc. Comput. Linguistics, 10:178–206, Feb. 2022. doi: 10.1162/tacl_a_00454 2, 4

    2022 doi

  15. [23]

    Hassan, F

    N. Hassan, F. Arslan, C. Li, and M. Tremayne. Toward automated fact- checking: Detecting check-worthy factual claims by ClaimBuster. InProc. ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining (KDD), KDD ’17, pp. 1803–1812. ACM, New York, Aug. 2017. doi: 10.1145/3097983. 3098131 2, 3

    2017 doi

  16. [24]

    T. L. Horstmann, B. Geisenberger, and M. Alam. T-REX: Table – refute or entail explainer. InProc. Eur. Conf. Mach. Learn. Principles Pract. Knowl. Discovery Databases (ECML PKDD), vol. 16022 ofLecture Notes in Computer Science, pp. 470–474. Springer, Porto, Portugal, Sep. 2025...

    2025 doi

  17. [25]

    Huang and A

    Y . Huang and A. Narechania. Facilitating proactive and reactive guidance for decision making on the web: A design probe with WebSeek. InProc. ACM CHI Conf. Human Factors Comput. Syst. (CHI), CHI ’26, art. no

  18. [26]

    K. Hyland. Stance and engagement: A model of interaction in academic discourse.Discourse Stud., 7(2):173–192, May 2005. doi: 10.1177/ 1461445605050365 2

    2005

  19. [27]

    Jahanbakhsh and D

    F. Jahanbakhsh and D. R. Karger. A browser extension for in-place signaling and assessment of misinformation. InProc. ACM CHI Conf. Human Factors Comput. Syst. (CHI), CHI ’24, art. no. 946. ACM, New York, May 2024. doi: 10.1145/3613904.3642473 2

    2024

  20. [28]

    S. M. Jones-Jang and Y . J. Park. How do people react to AI failure? automation bias, algorithmic aversion, and perceived controllability.J. Comput.-Mediated Commun., 28(1), art. no. zmac029, Jan. 2023. doi: 10. 1093/jcmc/zmac029 3

    2023

  21. [29]

    Juneja and T

    P. Juneja and T. Mitra. Human and technological infrastructures of fact- checking.Proc. ACM Hum.-Comput. Interact., 6(CSCW2), art. no. 418, Nov. 2022. doi: 10.1145/3555143 4

    2022 doi

  22. [30]

    Karagiannis, M

    G. Karagiannis, M. Saeed, P. Papotti, and I. Trummer. Scrutinizer: a mixed-initiative approach to large-scale, data-driven claim verification. Proc. VLDB Endowment, 13(12):2508–2521, Aug. 2020. doi: 10.14778/ 3407790.3407841 3

    2020

  23. [31]

    D. H. Kim, S. Choi, J. Kim, V . Setlur, and M. Agrawala. EmphasisChecker: A tool for guiding chart and caption emphasis.IEEE Trans. Visual. Comput. Graphics, 30(1):120–130, Jan. 2024. doi: 10.1109/TVCG.2023.3327150 2, 3, 8

    2024

  24. [32]

    D. H. Kim, E. Hoque, J. Kim, and M. Agrawala. Facilitating document reading by linking text and tables. InProc. ACM Symp. User Interface Softw. Technol. (UIST), UIST ’18, pp. 423–434. ACM, New York, Oct

  25. [33]

    H. Kim, K. D. Le, G. Lim, D. H. Kim, Y . J. Hong, and J. Kim. DataDive: Supporting readers’ contextualization of statistical statements with data exploration. InProc. ACM Int. Conf. Intelligent User Interfaces (IUI), IUI ’24, pp. 623–639. ACM, New York, Mar. 2024. doi: 10.1145...

    2024 doi

  26. [34]

    H.-P. Lee, A. Sarkar, L. Tankelevitch, I. Drosos, S. Rintel, R. Banks et al. The impact of generative AI on critical thinking: Self-reported reductions in cognitive effort and confidence effects from a survey of knowledge workers. InProc. ACM CHI Conf. Human Factors Comput. Sy...

    2025 doi

  27. [35]

    Lim and S

    G. Lim and S. T. Perrault. XAI in automated fact-checking? The benefits are modest and there’s no one-explanation-fits-all. InProc. Aust. Comput.- 10 © 2026 IEEE. This is the author’s version of the article that has been published in IEEE Transactions on Visualization and Comp...

    2026

  28. [36]

    C. Liu, Q. Zhou, X. Shen, X. B. Liu, T. Wu, and X. A. Chen. Behavioral indicators of overreliance during interaction with conversational language models. InProc. ACM CHI Conf. Human Factors Comput. Syst. (CHI), CHI ’26, art. no. 790. ACM, New York, Apr. 2026. doi: 10.1145/3772...

    2026 doi

  29. [37]

    Lloyd, T

    T. Lloyd, T. Nguyen, K. Levy, and M. Naaman. Beyond community notes: A framework for understanding and building crowdsourced context systems for social media. InProc. ACM CHI Conf. Human Factors Comput. Syst. (CHI), CHI ’26, art. no. 332. ACM, New York, Apr. 2026. doi: 10.1145...

    2026

  30. [38]

    Metropolitansky and J

    D. Metropolitansky and J. Larson. Towards effective extraction and eval- uation of factual claims. InProc. 63rd Annu. Meeting Assoc. Comput. Linguistics (ACL), pp. 6996–7045. Association for Computational Lin- guistics, Vienna, Austria, Jul. 2025. doi: 10.18653/v1/2025.acl-long.348 2

    2025 doi

  31. [39]

    Narechania, K

    A. Narechania, K. Odak, M. El-Assady, and A. Endert. ProvenanceWid- gets: A library of UI control elements to track and dynamically overlay analytic provenance.IEEE Trans. Visual. Comput. Graphics, 31(1):1235– 1245, Jan. 2025. doi: 10.1109/TVCG.2024.3456144 4

    2025

  32. [40]

    Narechania, A

    A. Narechania, A. Srinivasan, and J. Stasko. NL4DV: A toolkit for gener- ating analytic specifications for data visualization from natural language queries.IEEE Trans. Visual. Comput. Graphics, 27(2):369–379, Feb. 2021. doi: 10.1109/TVCG.2020.3030378 5

    2021

  33. [41]

    A. T. Nguyen, A. Kharosekar, S. Krishnan, S. Krishnan, E. Tate, B. C. Wallace et al. Believe it or not: Designing a human-AI partnership for mixed-initiative fact-checking. InProc. ACM Symp. User Interface Softw. Technol. (UIST), pp. 189–199. ACM, New York, Oct. 2018. doi: 10....

    2018

  34. [42]

    Y . Nie, H. Chen, and M. Bansal. Combining fact extraction and verification with neural semantic matching networks.Proc. AAAI Conf. Artif. Intell., 33(1):6859–6866, Jul. 2019. doi: 10.1609/aaai.v33i01.33016859 2

    2019 doi

  35. [43]

    Deaths involving COVID-19 by vac- cination status, England: deaths occurring between 2 January and 2 July 2021

    Office for National Statistics. Deaths involving COVID-19 by vac- cination status, England: deaths occurring between 2 January and 2 July 2021. ONS Statistical Bulletin, 2021. [Online]. Available: https://www.ons.gov.uk/peoplepopulationandcommunity/ birthsdeathsandmarriages/de...

    2021

  36. [44]

    Okamura and S

    K. Okamura and S. Yamada. Adaptive trust calibration for human-AI collaboration.PLoS ONE, 15(2), art. no. e0229132, Feb. 2020. doi: 10. 1371/journal.pone.0229132 3

    2020

  37. [45]

    OpenAI. GPT-4.1. OpenAI API model documentation, 2025. [Online]. Available: https://developers.openai.com/api/docs/models/ gpt-4.1. Accessed: Jul. 27, 2026. 4

    2025

  38. [46]

    Pesaranghader and E

    A. Pesaranghader and E. Li. Hallucination detection and mitigation in large language models, 2026. arXiv:2601.09929. [Online]. Available: https: //arxiv.org/abs/2601.09929. doi: 10.48550/arXiv.2601.09929 2, 4, 9

    2026 doi

  39. [47]

    Pirolli and S

    P. Pirolli and S. Card. The sensemaking process and leverage points for analyst technology as identified through cognitive task analysis. InProc. Int. Conf. Intelligence Analysis, vol. 5, pp. 2–4. McLean, May 2005. 3

    2005

  40. [48]

    Pirolli and D

    P. Pirolli and D. M. Russell. Introduction to this special issue on sense- making.Hum.–Comput. Interact., 26(1–2):1–8, Jan. 2011. doi: 10.1080/ 07370024.2011.556557 8

    2011

  41. [49]

    J. S. Rubinstein, D. E. Meyer, and J. E. Evans. Executive control of cognitive processes in task switching.J. Exp. Psychol.: Hum. Percept. Perform., 27(4):763–797, Aug. 2001. doi: 10.1037/0096-1523.27.4.763 4

    2001 doi

  42. [50]

    M. S. S. Samu, N. Khan, K. T. Elahi, T. B. Rahman, M. R. Islam, and F. Sadeque. AI as teammate or tool? a review of human–AI interaction in decision support, 2026. arXiv:2602.15865. [Online]. Available: https: //arxiv.org/abs/2602.15865. doi: 10.48550/arXiv.2602.15865 3

    2026 doi

  43. [51]

    D. E. P. Schultz. Truth goggles: automatic incorporation of context and primary source for a critical media experience. S.M. thesis, Program Media Arts Sci., School Archit. Planning, Massachusetts Inst. Technol., Cambridge, MA, USA, 2012. [Online]. Available: http://hdl.handle...

    2012

  44. [52]

    Sultanum and A

    N. Sultanum and A. Srinivasan. DataTales: Investigating the use of large language models for authoring data-driven articles. InProc. IEEE Visual- ization Visual Analytics (VIS), pp. 231–235. IEEE, Melbourne, Australia, Oct. 2023. doi: 10.1109/VIS54172.2023.00055 2, 3

    2023

  45. [53]

    Theologitis and D

    M. Theologitis and D. Suciu. Thucy: An LLM-based multi-agent sys- tem for claim verification across relational databases. InProc. AAAI Workshop LLM-Based Multi-Agent Syst. (LaMAS). Singapore, Jan. 2026. arXiv:2512.03278. doi: 10.48550/arXiv.2512.03278

    2026 doi

  46. [54]

    Thorne, A

    J. Thorne, A. Vlachos, C. Christodoulopoulos, and A. Mittal. FEVER: A large-scale dataset for fact extraction and VERification. InProc. Conf. North Amer. Chapter Assoc. Comput. Linguistics: Human Lang. Technol. (NAACL-HLT), pp. 809–819. Association for Computational Linguistic...

    2018 doi

  47. [55]

    Vlachos and S

    A. Vlachos and S. Riedel. Identification and verification of simple claims about statistical properties. InProc. Conf. Empirical Methods Natural Lang. Process. (EMNLP), pp. 2596–2601. Association for Computational Linguistics, Lisbon, Portugal, Sep. 2015. doi: 10.18653/v1/D15-1312

    2015 doi

  48. [56]

    Wadden, S

    D. Wadden, S. Lin, K. Lo, L. L. Wang, M. van Zuylen, A. Cohan et al. Fact or fiction: Verifying scientific claims. InProc. Conf. Empirical Methods Natural Lang. Process. (EMNLP), pp. 7534–7550. Association for Computational Linguistics, Online, Nov. 2020. doi: 10.18653/v1/2020...

    2020 doi

  49. [57]

    liar, liar pants on fire

    W. Y . Wang. “liar, liar pants on fire”: A new benchmark dataset for fake news detection. InProc. 55th Annu. Meeting Assoc. Comput. Lin- guistics (ACL), pp. 422–426. Association for Computational Linguistics, Vancouver, Canada, Jul. 2017. doi: 10.18653/v1/P17-2067 2

    2017 doi

  50. [58]

    Z. Wang, H. Zhang, C.-L. Li, J. M. Eisenschlos, V . Perot, Z. Wang et al. Chain-of-Table: Evolving tables in the reasoning chain for table understanding. InProc. Int. Conf. Learn. Representations (ICLR). Vienna, Austria, May 2024. doi: 10.48550/arXiv.2401.04398 2, 3

    2024 doi arXiv

  51. [59]

    Wongsuphasawat, D

    K. Wongsuphasawat, D. Moritz, A. Anand, J. Mackinlay, B. Howe, and J. Heer. V oyager: Exploratory analysis via faceted browsing of visualiza- tion recommendations.IEEE Trans. Visual. Comput. Graphics, 22(1):649– 658, Jan. 2016. doi: 10.1109/TVCG.2015.2467191 5

    2016

  52. [60]

    Wongsuphasawat, Z

    K. Wongsuphasawat, Z. Qu, D. Moritz, R. Chang, F. Ouk, A. Anand et al. V oyager 2: Augmenting visual analysis with partial view specifications. InProc. ACM CHI Conf. Human Factors Comput. Syst. (CHI), pp. 2648–

  53. [61]

    Y . Ye, B. Hui, M. Yang, B. Li, F. Huang, and Y . Li. Large language models are versatile decomposers: Decomposing evidence and questions for table- based reasoning. InProc. ACM Int. Conf. Res. Develop. Inf. Retrieval (SIGIR), pp. 174–184. ACM, New York, Jul. 2023. doi: 10.114...

    2023 doi

  54. [62]

    Z. Yuan, H. Yuan, C. Tan, W. Wang, and S. Huang. How well do large language models perform in arithmetic tasks?, 2023. arXiv:2304.02015. [Online]. Available: https://arxiv.org/abs/2304.02015. doi: 10. 48550/arXiv.2304.02015 2, 4, 9

    2023 doi arXiv

  55. [63]

    R. Zamith. Transparency, interactivity, diversity, and information prove- nance in everyday data journalism.Digit. Journalism, 7(4):470–489, Jan

  56. [800]

    ACM, New York, Apr. 2026. doi: 10.1145/3772318.3791945 4, 8

    2026

  57. [2014]

    doi: 10.1111/risa.12161 2

    doi

  58. [2019]

    doi: 10.1080/21670811.2018.1554409 1, 4, 9 11

    2018 arXiv

  59. [2024]

    doi: 10.1073/pnas.2322823121 4

    doi

  60. [2659]

    doi: 10.1145/3025453.3025768 5

    ACM, New York, May 2017. doi: 10.1145/3025453.3025768 5

    2017

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.