CoChat
Sign Up
Cited by 78Artificial Intelligence in EducationDigital literacy in educationInnovative Teaching and Learning Methods

Exploring How AI Literacy and Self-Regulated Learning Relate to Student Writing Performance and Well-Being in Generative AI-Supported Higher Education

Open access
Shi, Jiajia, Liu, Weitong, Hu, Ke·2025·Behavioral Sciences
DOI 10.3390/bs15050705PMC PMC12108985
Open this paper in CoChat — free accountOpen PDF

Abstract

The integration of generative artificial intelligence (GAI) into higher education is transforming students’ learning processes, academic performance, and psychological well-being. Despite the increasing adoption of GAI tools, the mechanisms through which students’ AI literacy and self-regulated learning (SRL) relate to their academic and emotional experiences remain underexplored. This study investigates how AI literacy and SRL are associated with writing performance and digital well-being among university students in GAI-supported higher learning contexts. A survey was administered to 257 students from universities in China, and structural equation modeling was used to examine the hypothesized relationships. Results show that both AI literacy and SRL significantly and positively predict students’ writing performance, with SRL having a stronger effect. Moreover, AI literacy shows a positive association with GAI-driven well-being, with writing performance serving as a partial mediator in this relationship. These findings suggest that fostering both technological competencies and effective learning strategies may support students’ academic outcomes while supporting their psychological well-being in AI-enriched educational environments. By integrating AI literacy and SRL into a unified model, this study contributes to the growing body of research on GAI-driven well-being in higher education and offers practical implications for cultivating balanced and sustainable learning experiences in the age of GAI.

AI snapshot

Written from the paper’s full text.

TL;DR

This study surveyed 257 university students in China and used structural equation modeling to examine how AI literacy and self-regulated learning relate to writing performance and GAI-driven well-being. Both AI literacy and SRL positively predicted writing performance, while AI literacy also positively related to well-being, with writing performance partially mediating that link.

Key findings

  • AI literacy (β = 0.153, p < 0.05) and SRL (β = 0.237, p < 0.01) were significantly positively associated with writing performance; SRL was the stronger predictor.
  • AI literacy was positively associated with GAI-driven well-being (β = 0.503, p < 0.001), and writing performance was also positively associated with GAI-driven well-being (β = 0.120, p < 0.01).
  • Writing performance partially mediated the AI literacy–well-being relationship (indirect β = 0.018) and fully mediated the SRL–well-being relationship (indirect β = 0.029).
  • Total effects on GAI-driven well-being were β = 0.521 for AI literacy, β = 0.120 for writing performance, and β = 0.029 for SRL.

Methods

  • A survey was administered to 257 valid undergraduate and graduate students from a high-level comprehensive university in northern China; 89 (34.6%) were male and 168 (65.3%) were female.
  • The questionnaire included demographic items and four multi-item 5-point Likert scales measuring AI literacy, SRL, writing performance, and GAI-driven well-being; data were collected online via Wenjuanxing from 12–18 December 2024.
  • Analyses used EFA in SPSS 22.0, CFA in AMOS 26, and SEM to test the hypothesized relationships; model fit was acceptable (χ²/df = 2.67, CFI = 0.91, IFI = 0.91, RMSEA = 0.08).

Limitations

  • The paper states that the model omitted potential control variables such as prior academic achievement, baseline digital literacy, access to GAI tools, and socioeconomic status.
  • The paper states that cross-sectional survey data restrict causal inference, and that the Chinese university sample may limit generalizability to other cultural and educational contexts.
  • Evident from the text: writing performance and well-being were measured through self-report Likert scales rather than objective writing or clinical well-being assessments, and the writing performance scale had the lowest reported reliability (Cronbach’s α = 0.68).

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3390/bs15050705

Cites

73 of 73 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 34 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Self-regulated learning was a stronger predictor of writing performance than AI literacy.

    “As Table 3 shows, the path analysis revealed that both AI literacy and SRL were positively associated with writing performance, with SRL being the stronger predictor.”
    • Also stated in: § 5.1. The Impact of AI Literacy and SRL on Writing Performance · boosted
      “The results demonstrated that both AI literacy and SRL were significantly and positively associated with writing performance, with SRL exerting a stronger effect.”
    Quote in the paperSubstantial claim
  • Findingown data

    AI literacy significantly and positively predicts university students' writing performance in GAI-supported higher learning contexts.

    “Results show that both AI literacy and SRL significantly and positively predict students’ writing performance, with SRL having a stronger effect.”
    Quote in the paperSubstantial claim
  • Findingown data

    The indirect effect from SRL to GAI-driven well-being through writing performance was significant (β = 0.029).

    “Similarly, SRL showed an indirect association with GAI-driven well-being through writing performance (β = 0.029).”
    Quote in the paperSubstantial claim
  • Findingown data

    Writing performance was positively associated with GAI-driven well-being (β = 0.120, p < 0.01).

    “Writing performance was also positively associated with GAI-driven well-being (β = 0.120, p < 0.01).”
    Quote in the paperSubstantial claim
  • Findingown data

    AI literacy was positively associated with GAI-driven well-being (β = 0.503, p < 0.001).

    “Furthermore, AI literacy was positively associated with GAI-driven well-being (β = 0.503, p < 0.001), indicating that learners with higher AI literacy tended to report higher digital well-being in GAI-supported environments.”
    • Also stated in: § 5.2. The Impact of AI Literacy on GAI-Driven Well-Being · plain
      “The findings also revealed that AI literacy was positively associated not only with writing performance but also with students’ GAI-driven well-being.”
    Quote in the paperSubstantial claim
  • May combine two claims. 19
  • Classified as probably not a substantial claim. 20

  • May combine two claims.

    Findingown data

    AI literacy had the largest standardized total effect on GAI-driven well-being (β = 0.521), followed by writing performance (β = 0.120) and SRL (β = 0.029).

    “AI literacy emerged as the most influential predictor (β = 0.521), followed by writing performance (β = 0.120) and SRL (β = 0.029).”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    SRL theory posits that students with higher self-regulatory capacities are more adept at employing metacognitive strategies and adjusting their learning behaviors to improve performance.

    “Moreover, the findings are consistent with SRL theory (Zimmerman, 2002), which posits that students with higher self-regulatory capacities are more adept at employing metacognitive strategies and adjusting their learning behaviors to improve performance.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Zimmerman, 2002 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    Data were collected using random sampling procedures.

    “Data were collected over a one-week period using random sampling procedures.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    SRL remains a critical antecedent of performance-related gains, which in turn is associated with better well-being outcomes in GAI-mediated learning contexts.

    “Furthermore, the findings suggest that while SRL may not directly affect GAI-driven well-being, it remains a critical antecedent of performance-related gains, which in turn is associated with better well-being outcomes in GAI-mediated learning contexts.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The mediation model assumes that AI literacy and SRL precede writing performance, which precedes GAI-driven well-being.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    SRL indirectly improves writing outcomes by strengthening learners' language-specific self-efficacy.

    “Moreover, SRL has been found to indirectly improve writing outcomes by strengthening learners’ language-specific self-efficacy (Teng & Zhang, 2020).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Teng & Zhang, 2020

      Not checked in the free run.

  • May combine two claims.

    Study design and conductcitation

    The final sample of 257 participants exceeds the recommended minimum of 100 to 150 for SEM, supporting the robustness of the SEM analysis.

    “To ensure the adequacy of the sample size for SEM, prior research suggests a minimum of 100 to 150 participants (Kline, 2010). With a final sample of 257 participants, the dataset exceeds the recommended threshold, supporting the robustness of the SEM analysis.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Kline, 2010

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

    Interpretationown data

    The findings provide empirical support to key assumptions derived from existing theories and extend the application of relevant theoretical frameworks in the domain of GAI-supported learning.

    “The findings provide empirical support to key assumptions derived from existing theories and extend the application of relevant theoretical frameworks in the domain of GAI-supported learning.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    The results support the hypothesized model and demonstrate that both AI literacy and SRL strategies are important correlates of students' writing performance and well-being in GAI-assisted learning.

    “These results support the hypothesized model and demonstrate that both AI literacy and SRL strategies are important correlates of students’ writing performance and well-being in the context of GAI-assisted learning.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Students who are proficient in using AI tools benefit from enhanced writing quality and reduced writing anxiety and tend to experience a stronger sense of control over the learning process.

    “The results of the present study further suggest that students who are proficient in using AI tools not only benefit from enhanced writing quality and reduced writing anxiety but also tend to experience a stronger sense of control over the learning process.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductown data

    The questionnaire items were pilot-tested before the formal survey and demonstrated satisfactory reliability and construct validity.

    “The items were pilot-tested prior to the formal survey and demonstrated satisfactory reliability and construct validity.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Student data used by GAI systems present risks of misuse or privacy breaches that may compromise students' psychological and emotional well-being.

    “While such data are often leveraged to optimize learning outcomes, they also present risks of misuse or privacy breaches, potentially compromising students’ psychological and emotional well-being (Malik et al., 2024).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Malik et al., 2024

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Students may become more socially isolated and less capable of engaging in in-person communication and collaboration as a result of excessive GAI reliance.

    “As a result, students may become more socially isolated and less capable of engaging in in-person communication and collaboration, factors that are crucial to their broader social well-being and developmental outcomes (Cambra-Fierro et al., 2025; Rodway & Schepman, 2023; Zhai et al., 2024).”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Cambra-Fierro et al., 2025

      Not checked in the free run.

    • cites Rodway & Schepman, 2023

      Not checked in the free run.

    • cites Zhai et al., 2024 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    GAI can save time and offer personalized feedback and automated evaluation.

    “It can also save time and offer personalized feedback and automated evaluation (Lund & Wang, 2023; Chen, 2023; Salvagno et al., 2023; Zawacki-Richter et al., 2019).”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Lund & Wang, 2023 source

      Not checked in the free run.

    • cites Chen, 2023 source

      Not checked in the free run.

    • cites Salvagno et al., 2023 source

      Not checked in the free run.

  • May combine two claims.

    Study design and conductown data

    The average variance extracted (AVE) for each construct ranged from 0.51 to 0.74 and exceeded the cut-off value of 0.5, demonstrating adequate construct validity.

    “The AVE of each construct ranged from 0.51 to 0.74 and exceeded the cut-off value of 0.5, demonstrating adequate construct validity because more than 50% variance is explained by the construct.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductown data

    The measurement model showed satisfactory internal reliability and convergent and discriminant validity.

    “The results indicate satisfactory internal reliability and convergent and discriminant validity.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductown data

    The composite reliability (CR) for each construct ranged from 0.76 to 0.91, all above the recommended threshold of 0.60.

    “The CR for each construct ranged from 0.76 to 0.91, all higher than the recommended threshold of 0.60.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    GAI systems typically require access to large volumes of student data, including academic records and personal information.

    “GAI systems typically require access to large volumes of student data, including academic records and even personal information (Akgun & Greenhow, 2022).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Akgun & Greenhow, 2022 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationreasoning

    The omission of these factors may introduce bias and limit the precision of the estimated relationships.

    “The omission of these factors may introduce bias and limit the precision of the estimated relationships.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductreasoning

    Feedback from the two pilot-test groups was used to refine the questionnaire items and improve construct validity.

    “Feedback from both groups was used to refine the questionnaire items and improve construct validity.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    GAI can improve writing quality by enhancing grammar, vocabulary, and fluency.

    “Recent studies show that GAI can improve writing quality by enhancing grammar, vocabulary, and fluency.”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Lund & Wang, 2023 source

      Not checked in the free run.

    • cites Chen, 2023 source

      Not checked in the free run.

    • cites Salvagno et al., 2023 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    GAI may pose risks to students' psychological well-being due to technology dependency, ethical concerns, and reduced social connectedness.

    “it may also pose risks to students’ psychological well-being due to technology dependency, ethical concerns, and reduced social connectedness (Crawford et al., 2024).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Crawford et al., 2024 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    GAI can reduce instructional pressure, increase engagement, empathy development, and promotion of well-being through personalized support.

    “while GAI can reduce instructional pressure, increase engagement, empathy development, and promotion of well-being (Sethi & Jain, 2024) through personalized support”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Sethi & Jain, 2024

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Targeted SRL interventions can significantly enhance students' writing performance.

    “Empirical studies have shown that targeted SRL interventions can significantly enhance students’ writing performance (Bai, 2015; De Silva & Graham, 2015).”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Bai, 2015

      Not checked in the free run.

    • cites De Silva & Graham, 2015

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    SRL enables learners to actively monitor and control their cognitive engagement, which is particularly important in complex tasks such as academic writing.

    “Pintrich (2000) further emphasized that SRL enables learners to actively monitor and control their cognitive engagement, which is particularly important in complex tasks such as academic writing.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Pintrich (2000)

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Interactions with GAI tools in emotionally expressive or supportive contexts may contribute directly to improved affective states.

    “Recent research further suggests that interactions with GAI tools themselves—especially in emotionally expressive or supportive contexts—may also contribute directly to improved affective states (Hu et al., 2025), expanding the potential impact of GAI beyond academic outcomes.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Hu et al., 2025

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Overdependence on GAI may lead to digital fatigue, loneliness, anxiety, and technostress.

    “However, overdependence on GAI may also lead to digital fatigue, loneliness, anxiety, and technostress (Cambra-Fierro et al., 2025; Crawford et al., 2024; Klimova & Pikhart, 2025).”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Cambra-Fierro et al., 2025

      Not checked in the free run.

    • cites Crawford et al., 2024 source

      Not checked in the free run.

    • cites Klimova & Pikhart, 2025

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Students viewed AI tools as collaborative partners-helping them brainstorm ideas, foster creativity, and reduce feelings of isolation during writing tasks.

    “Furthermore, J. Kim and Cho (2023) found that students viewed AI tools as collaborative partners—helping them brainstorm ideas, foster creativity, and reduce feelings of isolation during writing tasks.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites J. Kim and Cho (2023)

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    AI technologies can improve engagement and satisfaction by offering personalized learning paths, real-time feedback, and intelligent recommendation systems.

    “For example, Luckin et al. (2016) noted that AI technologies can improve engagement and satisfaction by offering personalized learning paths, real-time feedback, and intelligent recommendation systems.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Luckin et al. (2016)

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisecitation

    Positive achievement emotions are expected to relate to better academic outcomes across a variety of learning tasks.

    “These positive emotions, in turn, are expected to relate to better academic outcomes across a variety of learning tasks.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Pekrun, 2006; Pekrun et al., 2014

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    AI support improves writing structure and contributes to a more enjoyable learning experience.

    “M. P.-C. Lin and Chang (2020) similarly reported that AI support improves writing structure and contributes to a more enjoyable learning experience.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites M. P.-C. Lin and Chang (2020)

      Not checked in the free run.

30 of 30

  • The paper relies on this without support.

    Few studies have built a model integrating AI literacy, SRL, writing performance, and well-being.

    “Although past research has examined the link between student traits, performance, and well-being in digital environments, a few studies have built a model integrating all these factors.”

  • The paper relies on this without support.

    The interrelationships among AI literacy, SRL, writing performance, and GAI-driven well-being have not been systematically examined.

    “To date, the interrelationships among AI literacy, SRL, writing performance, and GAI-driven well-being have not been systematically examined.”

  • The paper relies on this without support.

    Existing research tends to treat AI literacy, SRL, writing performance, and well-being in isolation or through linear models, with limited attention to how they might function synergistically.

    “Moreover, existing research tends to treat these variables in isolation or through linear models, with limited attention to how they might function synergistically in AI-enhanced learning processes.”

  • The paper relies on this without support.

    The mediation model assumes that AI literacy and SRL precede writing performance, which precedes GAI-driven well-being.

    “Although the mediation model was theoretically informed, the findings reflect correlational patterns rather than confirmed temporal or causal effects.”
4 of 4

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?

  • Undergraduates prioritize efficiency and accessibility, while postgraduates express ethical ambiguity and concerns about voice erosion despite recognizing benefits for lexical diversity.

    “Undergraduates prioritize efficiency and accessibility, while postgraduates express ethical ambiguity and concerns about voice erosion despite recognizing benefits for lexical diversity (Hanh and Duyen, 2025; Shi et al., 2025).”

    From Reimagining writing assessment for the AI era: a systematic review on balancing AI support and authentic skill growth (2026) · cites this paper as Shi et al., 2025

    The Introduction says generative AI can save time and improve vocabulary, and the Literature Review (§2.2) discusses ethical concerns about AI-assisted writing. The Methods (§3.1) include both undergraduate and graduate students, but the passages do not report the claimed difference in their priorities or concerns, nor postgraduate concerns about voice erosion or recognition of lexical diversity. (LLM, conf=0.81)

    Experimental — accuracy not yet measured

  • Publication bias assessment revealed a "positive-utility" reporting bias favoring successful implementations while null or negative findings remain scarce.

    “Following PRISMA 2020 guidelines, publication bias assessment revealed a “positive-utility” reporting bias favoring successful implementations while null or negative findings remain scarce (Kumar et al., 2025; Shi et al., 2025).”

    From Reimagining writing assessment for the AI era: a systematic review on balancing AI support and authentic skill growth (2026) · cites this paper as Shi et al., 2025

    The Statistical Analyses section describes factor analysis and structural equation modeling, not a PRISMA review or publication-bias assessment. The Limitations section discusses possible bias from omitted variables, but no passage found reports a “positive-utility” reporting bias or a scarcity of null or negative findings (searched: positive-utility reporting bias; PRISMA 2020 publication bias assessment). (LLM, conf=0.76)

    Experimental — accuracy not yet measured

Self-check

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

Each contradiction between two passages listed here was confirmed by an independent check.

  • Abstract describes multiple universities, but methods describe one

    Minor

    The abstract says students were surveyed from universities in China, while the methods state that participants came from a single university.

1 of 1

Cites

Flagged first, then by citations

Research works printed but not linked: 4

  1. Banovac, 2023, ChatGPT-3.5 as writing assistance in students’ essays, Humanities and Social Sciences Communications, 10, 1
  2. Hair, J. F., Black, W. C., Babin, B. J., and Anderson, R. E. (2014). Multivariate data analysis, Pearson. [7th ed.].
  3. Kline, R. B. (2010). Principles and practice of structural equation modeling, Guilford Press. [3rd ed.].
  4. Rowland, 2023, Two frameworks to guide discussions around levels of acceptable use of generative AI in student academic research and writing, Journal of Academic Language and Learning, 17, T31

Cited by

Sorted by citations

Related

Authors
Shi, J., Liu, W., & Hu, K.
Affiliations
  • Shandong University (CN)
  • Fujian Normal University (CN)
Corresponding author
  • Liu, Weitong
  • Hu, Ke
ORCID
Venue
Behavioral Sciences
Publisher
MDPI AG
Published
2025-05-20
Volume
15
Issue
5
Pages
705
ISSN
2076-328X (ISSN-L)
DOI
10.3390/bs15050705
PMC
PMC12108985
OpenAlex
W4410539449
Cited by
77
References
71
Citations by year
2026: 72 · 2025: 5
FWCI
80.06
Type
journal article
Language
en
Open access
Licence
Locations
  • Behavioral Sciences (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
National Education Science Planning General Project “Research on Digital Modeling and Assessment of College Students’ Learning Processes Based on Multimodal Data” (BIA240130)
Topics
  • Physical Sciences › Computer Science › Information Systems › Artificial Intelligence in Education
  • Physical Sciences › Computer Science › Information Systems › Digital literacy in education
  • Social Sciences › Psychology › Developmental and Educational Psychology › Innovative Teaching and Learning Methods
Keywords
AI literacy, student self-regulation, generative AI, student writing performance, digital well-being, university students, higher education, AI-assisted learning

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.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-06.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-06.

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