Pith. sign in

REVIEW 3 major objections 5 minor 120 references

The format of an AI disclosure is not neutral: how the collaboration is drawn shifts whether readers see an article as human- or AI-written.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

Disclosure visualization format systematically shifts readers' perceptions of human vs AI contribution: role-based timelines amplify perceived AI role in mostly human articles, while task-based timelines make mostly AI articles seem more human.

T0 review reviewed 2026-08-03 challenge →

load-bearing objection Novel finding on disclosure format bias, but the reported CIs undercut the main effect and the prototype content is not fully faithful to the manipulation; still deserves peer review. the 3 major comments →

arxiv 2601.11072 v1 pith:ZLO2GKON submitted 2026-01-16 cs.HC cs.AIcs.CY

More Human or More AI? Visualizing Human-AI Collaboration Disclosures in Journalistic News Production

Amber Kusters , Pooja Prajod , Pablo Cesar , Abdallah El Ali This is my paper
classification cs.HC cs.AIcs.CY
keywords AI disclosurehuman-AI collaborationjournalisminformation visualizationtransparencyperceptioneye trackingnews production
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

News outlets currently disclose AI involvement with simple labels, which tell readers that AI was used but not how. This paper asks whether the visual form of a disclosure changes what readers conclude about who wrote an article. Through co-design sessions and a lab study with eye tracking and interviews, the authors built four disclosure prototypes—a text label, a role-based timeline, a chatbot, and a task-based timeline—and showed readers the same articles with different collaboration ratios. Their core finding is that the visualization format itself shifts perception: role-based timelines make readers attribute more work to AI in mostly human-written articles, while task-based timelines make readers see more human involvement in mostly AI-written articles. If this holds, disclosure design is a form of framing, not a neutral conduit, and newsrooms' choice of disclosure graphic will partly determine how readers judge authorship.

Core claim

The paper's central claim is that human-AI collaboration disclosures in journalism are not passive carriers of information: the visual structure of a disclosure systematically alters how readers balance human versus AI credit. In a within-subjects experiment (N=32) with four prototypes derived from 69 co-designed concepts, all formats communicated the high/low AI ratio, but directionally different formats skewed perception. The Role-based Timeline—a linear row of person/robot icons—amplified perceived AI contribution in primarily human-written articles (significant against text, chatbot, and task timeline), while the Task-based Timeline—a five-stage workflow with hover details—shifted percep

What carries the argument

The load-bearing objects are the four disclosure prototypes, particularly the two timeline designs. The Role-based Timeline shows a linear sequence of who did what (robot vs person icons with captions such as 'ChatGPT-4o researched and gathered data'), while the Task-based Timeline shows the five editorial stages (ideation, research, writing, headline, review/publish) with human/AI tags and hover-revealed detail. The contrast between 'who worked' (role framing) and 'what was done' (task framing) is the mechanism that produces asymmetric perception shifts. Supporting machinery includes the co-design session analysis that yielded 69 designs, a selection procedure based on HCI heuristics (simpl

Load-bearing premise

The timeline disclosures depict workflow steps (e.g., 'ChatGPT-4o researched and gathered data') that were not actually logged from the production of the stimulus articles; if those steps are arbitrary or inaccurate, the observed perception shifts could be an artifact of the invented narrative rather than of the visualization format.

What would settle it

Repeat the study holding the production ground truth fixed—e.g., instrument an actual editorial workflow to log each contribution—and generate disclosures strictly from that log. If the role-based amplification and task-based humanization effects disappear when the steps are faithful, the claim that format alone shifts perception fails. Alternatively, present the same two article texts with identical step content but permuted iconography (robot-first vs person-first) and check whether the skew reverses.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If disclosure visualizations actively frame authorship, then the choice of visualization is an editorial decision with measurable consequences for reader trust and attribution, not a mere formatting detail.
  • Text-only disclosures, the current common practice, are the least effective at communicating human-AI collaboration, so regulatory compliance via simple labels may fail its transparency purpose.
  • Role-based timelines risk overstating AI's role in primarily human articles, which could deter readers or unfairly diminish journalist credit; task-based timelines risk overstating human involvement in primarily AI articles, which could hide AI's actual contribution.
  • Interactive formats like the chatbot buy depth at the cost of overview and comprehension, making them better suited to high-stakes articles where readers are willing to invest time.
  • The pattern suggests disclosures should be matched to article type and stakes rather than applied uniformly, and that consistency matters more than personalization to avoid misreading.

Where Pith is reading between the lines

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

  • Because the prototypes' step-by-step narratives (e.g., 'ChatGPT-4o researched and gathered data') were invented for the stimuli rather than logged from an actual editorial workflow, the observed perception shifts may be driven by the specific story told in the timeline rather than by the role-vs-task format per se. A follow-up with a ground-truth step log would disentangle format from content.
  • The fact-checking effect hints at a general principle: readers anchor AI attribution to tasks they already associate with automation (verification, data gathering). Disclosure designers might exploit or correct for such anchors depending on transparency goals.
  • If disclosures are framed by the same GenAI tools they describe, the disclosure itself becomes a trust artifact; the paper's caution about AI-generated disclosures suggests a need for independent provenance verification, which could be tested by asking readers to evaluate the same timeline when produced by a third party vs by the AI.
  • The authors' result that higher AI literacy reduces perceived AI involvement in headline and topic selection while ChatGPT experience increases it suggests that 'transparency' is filtered through prior beliefs; public AI-literacy interventions might change how any given disclosure lands.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper designs and evaluates four visualizations for disclosing human–AI collaboration in news articles: Textual Disclosure, Role-based Timeline, Chatbot, and Task-based Timeline. The authors first ran co-design sessions (N=10) that produced 69 design concepts, from which four prototypes were built. They then ran a within-subjects lab study (N=32; eye-tracking subset N=20) with two collaboration-ratio conditions (Primarily Human vs. Primarily AI) and measured perceived collaboration, perceived AI roles, clarity, informativeness, gaze patterns, and qualitative preferences. The headline findings are that textual disclosures were least effective, the Chatbot provided the most in-depth information, the timelines gave clearer overviews, and—most centrally—that role-based timelines amplified perceived AI contribution in primarily human articles while task-based timelines shifted primarily AI articles toward perceived human involvement. The paper argues that disclosure visualizations are not neutral conduits but can actively reframe perceived authorship.

Significance. If the central comparative claim holds, this is a useful contribution to the growing literature on AI disclosure and human–AI collaboration transparency. The co-design process is well documented, the four prototypes cover a reasonable design space, and the mixed-methods evaluation combines questionnaires, eye tracking, and interviews. Statistical modeling with cumulative link mixed-effects models and FDR correction is appropriate for the ordinal and repeated-measures data. The qualitative data add texture and help explain quantitative patterns. The paper also gives concrete design considerations and acknowledges several limitations. The strongest value is the cautionary point that disclosure format may change readers' perception of authorship, not merely inform them—but this claim currently rests on a format/content confound that needs to be resolved or substantially reworded before the paper can be accepted.

major comments (3)
  1. [§4.3.2 and Fig. 1] The central claim (Abstract; §4.4.2) that visualization format systematically changes perceived human–AI balance is undermined by a format/content confound. The stimulus manipulation varied only (1) headline/style generation from a human article vs (2) full article generation from a headline. Yet the timeline prototypes display a five-stage workflow including 'ChatGPT-4o researched and gathered data' and 'Journalist came up with the idea'—tasks that appear in no prompt. Thus the timeline conditions differ from Textual/Chatbot in asserted semantic content, not only visual encoding. The observed amplification (e.g., RT increasing AI perception in human articles) may be caused by the claim that AI performed research, not by the timeline format. The sentence in §4.3.2 that prototype information 'was derived from these controlled prompt-based manipulations' does not address this; a mapping fr
  2. [§5.4 and §4.4.8] The paper acknowledges an intentional confound in information granularity: the Chatbot could retrieve more details and the timelines show more steps. This is more serious than a limitation. The claims 'Chatbot offered the most in-depth information' and 'Timelines provided clearer overviews of editorial steps' are, to a substantial degree, restatements of the design: the Chatbot contained more text and the timelines contained step labels. Since the prototypes were not equated on information content or amount, RQ2 cannot separate format effects from content effects. Please either add a control condition that holds content constant across formats or downgrade the causal wording throughout the abstract and §5.2.
  3. [§4.3 Hardware/Software and §4.4.10–14] The eye-tracking analyses rest on N=20 after a post-hoc increase of the accuracy threshold from Tobii's 0.8° to 1.6°. With 12 of 32 participants excluded, this is a substantial selection step, and no sensitivity analysis is reported. The very large effect sizes (partial η² = .37–.71) make it plausible the qualitative gaze conclusions survive, but the post-hoc threshold should be justified and a full-sample or threshold-robustness analysis reported. At minimum, the paper should state this as a limitation in §5.4 and soften the gaze-based claims.
minor comments (5)
  1. [§4.2.1] Typo: '32 participants 4 (19 female, 12 male, 1 non-binary)' should read '32 participants (19 female, 12 male, 1 non-binary)'.
  2. [Fig. 13] The significance bars contain garbled asterisk strings (e.g., '*** ******', '*********') that appear to be rendering artifacts. Please clean the figure so the intended significance levels are legible.
  3. [§4.3.2] Grammar: 'The information represented in these prototypes were derived' should be 'was derived'.
  4. [General] The header contains inconsistent metadata: '©2018', 'Manuscript submitted to ACM', and 'Preprint - Accepted to ACM CHI 2026' all appear. This should be standardized before publication.
  5. [§4.4.2 / Fig. 10] The labels 'True collaboration' and 'Perceived collaboration' in Fig. 10 are confusing; clarify which axis corresponds to the experimental condition and which to the participant response.

Circularity Check

0 steps flagged

No significant circularity: the central findings are empirical comparisons from a controlled study, not derivations from fitted inputs or self-citation chains.

full rationale

This paper is an empirical HCI study rather than a derivation chain. The main claims—role-based timelines amplify perceived AI contribution in primarily human articles and task-based timelines shift primarily AI articles toward human involvement—come from a within-subjects lab study with CLMM/LMM analyses of questionnaire and eye-tracking data (Sec. 4.4.2). The independent variables (visualization type, collaboration ratio) and dependent variables (perceived collaboration, AI-role ratings, gaze metrics) are distinct, and no parameter is fitted to an outcome and then reported as a prediction. Self-citations ([22], [28], [66], [111]) support motivation and general framing but are not load-bearing for the empirical results; there is no imported uniqueness theorem, no ansatz smuggled via citation, and no renaming of a known result. The closest concern is the design confound the authors themselves acknowledge in Sec. 5.4: the prototypes intentionally differed in information granularity, so the finding that the Chatbot offers the most in-depth information may partly reflect the amount of content embedded in the stimulus. That is an internal-validity limitation, not a circular derivation: the outcome was measured, not assumed, and the paper explicitly flags the confound rather than disguising it as a prediction. A confound of this kind does not reduce the central format-effect findings to the inputs by construction. Therefore, under the stated rules requiring an exhibited reduction, there is no significant circularity and the score is 0.

Axiom & Free-Parameter Ledger

2 free parameters · 4 axioms · 0 invented entities

The paper is an empirical HCI study: no mathematical derivation, no new physical entities. The main 'free' choices are data-processing thresholds and the author-designed prototype content; the main axioms are the validity of the collaboration-ratio manipulation and the representativeness of the chosen prototypes.

free parameters (2)
  • Eye gaze validity threshold = 80% valid gaze points
    Data inclusion criterion; led to exclusion of 8 participants.
  • Eye tracker accuracy threshold = 1.6° (Tobii default 0.8°)
    Adjusted upward 'due to the large Areas of Interest'; led to final N=20 for gaze data.
axioms (4)
  • domain assumption The four prototypes selected by two authors are representative of the 69 co-designed concepts
    Selection criteria in Table 1 are qualitative and applied by the authors; may not cover the full design space.
  • domain assumption The 'Primarily Human' vs 'Primarily AI' article versions are valid operationalizations of collaboration ratios
    Human version only uses ChatGPT for headline/style from a human article; AI version uses ChatGPT to write a 400-word article from a headline (Sec 4.3.2).
  • ad hoc to paper The process narrative in the prototypes (e.g., 'ChatGPT-4o researched and gathered data') matches the actual workflow of the stimulus articles
    The actual manipulation does not include a research/data-gathering step; the visualizations' workflow is authored by the researchers, not recorded.
  • domain assumption Gaze metrics are valid proxies for information processing/understanding
    Standard eye-tracking assumption; not independently validated in this study.

reviewed 2026-08-03 · how reviews work

0 comments
Cite this review

Pith. "Pith review of More Human or More AI? Visualizing Human-AI Collaboration Disclosures in Journalistic News Production." pith.science (2026). https://pith.science/paper/ZLO2GKON

@misc{pith2026260111072,
  author       = {Pith},
  title        = {Pith review of: More Human or More AI? Visualizing Human-AI Collaboration Disclosures in Journalistic News Production},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZLO2GKON}},
  note         = {Machine review of arXiv:2601.11072}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Within journalistic editorial processes, disclosing AI usage is currently limited to simplistic labels, which misses the nuance of how humans and AI collaborated on a news article. Through co-design sessions (N=10), we elicited 69 disclosure designs and implemented four prototypes that visually disclose human-AI collaboration in journalism. We then ran a within-subjects lab study (N=32) to examine how disclosure visualizations (Textual, Role-based Timeline, Task-based Timeline, Chatbot) and collaboration ratios (Primarily Human vs. Primarily AI) influenced visualization perceptions, gaze patterns, and post-experience responses. We found that textual disclosures were least effective in communicating human-AI collaboration, whereas Chatbot offered the most in-depth information. Furthermore, while role-based timelines amplified AI contribution in primarily human articles, task-based timeline shifted perceptions toward human involvement in primarily AI articles. We contribute Human-AI collaboration disclosure visualizations and their evaluation, and cautionary considerations on how visualizations can alter perceptions of AI's actual role during news article creation.

Figures

Figures reproduced from arXiv: 2601.11072 by Abdallah El Ali, Amber Kusters, Pablo Cesar, Pooja Prajod.

Figure 1
Figure 1. Figure 1: Our human–AI collaboration disclosure visualizations, with an example for a primarily AI-written article. From top to bottom: [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The two part study approach with contributions outlined in (bold) blue. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Screenshot of a filled out sensitizing booklet in Miro, as part of our preparation for the co-design session. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: In-person co-design session setup. Visualization techniques to inspire, but not restrict them. During the ideation rounds they received ideation templates ( [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Example ideation template for co-design session [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Two examples of the results of the co-design sessions, drawn on the ideation template. The example on the left was created [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: User study procedure. The four Articles refer to the four article versions (1-Human, 1-AI, 2-Human, 2-AI), counterbalanced [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: User study setup All interviews were audio-recorded for later transcription and thematic analysis. Participants were compensated with a $/€10 gift card. 4.2.1 Participants. 32 participants4 (19 female, 12 male, 1 non-binary) were recruited through multiple platforms and snowball sampling. Ages ranged from 22 to 72. Educational backgrounds were varied (completed high school, bachelor, master, and doctorate)… view at source ↗
Figure 9
Figure 9. Figure 9: Heatmap of the gaze points of one participant on a stimulus. The red boxes indicate the predefined AOIs. [PITH_FULL_IMAGE:figures/full_fig_p014_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Confusion matrix of human-AI collaboration per article [PITH_FULL_IMAGE:figures/full_fig_p015_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Proportion of responses of perceived human-AI collaboration per article and disclosure visualization [PITH_FULL_IMAGE:figures/full_fig_p016_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Perceived AI involvement per role by visualization and collaboration ratio. Significance bar in black refer to significant main [PITH_FULL_IMAGE:figures/full_fig_p018_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Combined plots of response variables and gaze features per disclosure visualization. [PITH_FULL_IMAGE:figures/full_fig_p020_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Task-based Timeline to disclose our AI usage. [PITH_FULL_IMAGE:figures/full_fig_p028_14.png] view at source ↗

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

120 extracted references · 12 canonical work pages · 2 internal anchors

  1. [1]

    2022.Measuring the user experience: Collecting, analyzing, and presenting UX metrics

    Bill Albert and Tom Tullis. 2022.Measuring the user experience: Collecting, analyzing, and presenting UX metrics. Morgan Kaufmann

    2022

  2. [2]

    Sacha Altay and Fabrizio Gilardi. 2024. People are skeptical of headlines labeled as AI-generated, even if true or human-made, because they assume full AI automation.PNAS nexus3, 10 (2024), pgae403

    2024

  3. [3]

    Agnieszka Andrychowicz-Trojanowska. 2018. Basic terminology of eye-tracking research.Applied Linguistics Papers25/2 (2018), 123–132

    2018

  4. [4]

    Theo Araujo, Anna Brosius, Andreas C Goldberg, Judith Möller, and Claes de Vreese. 2023. Humans vs. AI: the role of trust, political attitudes, and individual characteristics on perceptions about automated decision making across Europe.International Journal of Communication17 (2023), 28

    2023

  5. [5]

    Douglas Bates, Martin Mächler, Ben Bolker, and Steve Walker. 2015. Fitting Linear Mixed-Effects Models Using lme4.Journal of Statistical Software 67, 1 (2015), 1–48. https://doi.org/10.18637/jss.v067.i01

    doi:10.18637/jss.v067.i01 2015

  6. [6]

    Elisa Bertino, Finale Doshi-Velez, Maria Gini, Daniel Lopresti, and David Parkes. 2020. Artificial Intelligence & Cooperation.arXiv e-prints, Article arXiv:2012.06034 (Dec. 2020), arXiv:2012.06034 pages. https://doi.org/10.48550/arXiv.2012.06034 arXiv:2012.06034 [cs.CY]

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2012.06034 2020

  7. [7]

    2024.Proportions

    Francesco Bianconi. 2024.Proportions. Springer Nature Switzerland, Cham, 25–52. https://doi.org/10.1007/978-3-031-57051-3_3

    work page doi:10.1007/978-3-031-57051-3_3 2024

  8. [8]

    Monika Bickert. 2024. Our Approach to Labeling AI-Generated Content and Manipulated Media. https://about.fb.com/news/2024/04/metas- approach-to-labeling-ai-generated-content-and-manipulated-media/ Accessed: 2025-01-27

    2024

  9. [9]

    Abeba Birhane, William Isaac, Vinodkumar Prabhakaran, Mark Diaz, Madeleine Clare Elish, Iason Gabriel, and Shakir Mohamed. 2022. Power to the People? Opportunities and Challenges for Participatory AI. InEquity and Access in Algorithms, Mechanisms, and Optimization(Arlington, VA, USA)(EAAMO ’22). Association for Computing Machinery, New York, NY, USA, Arti...

    arXiv 2022

  10. [10]

    Adam Block, Ayush Sekhari, and Alexander Rakhlin. 2025. GaussMark: A Practical Approach for Structural Watermarking of Language Models. arXiv:2501.13941 [cs.CR] https://arxiv.org/abs/2501.13941

    Pith/arXiv arXiv 2025

  11. [11]

    Michał Boni. 2021. The ethical dimension of human–artificial intelligence collaboration.European View20, 2 (2021), 182–190

    2021

  12. [12]

    Virginia Braun and Victoria Clarke. 2012. Thematic analysis. InAPA Handbook of Research Methods in Psychology, Vol. 2: Research Designs: Quantitative, Qualitative, Neuropsychological, and Biological, Harris Cooper, Paul M. Camic, Deborah L. Long, A. T. Panter, David Rindskopf, and Kenneth J. Sher (Eds.). American Psychological Association, 57–71. https://...

    doi:10.1037/13620-004 2012

  13. [13]

    Karin Breckner, Thomas Neumayr, Martina Mara, Marc Streit, and Mirjam Augstein. 2025. The Changing Nature of Human-AI Relations: A Scoping Review on Terminology and Evolvement in the Scientific Literature.International Journal of Human–Computer Interaction0, 0 (2025), 1–58. https://doi.org/10.1080/10447318.2025.2482742 arXiv:https://doi.org/10.1080/104473...

    arXiv 2025

  14. [14]

    Peter Bro, Kenneth Reinecke Hansen, and Ralf Andersson. 2016. Improving productivity in the newsroom? Deskilling, reskilling and multiskilling in the news media.Journalism Practice10, 8 (2016), 1005–1018. 28 Visualizing Human-AI Collaboration Disclosures Woodstock ’18, June 03–05, 2018, Woodstock, NY

    2016

  15. [15]

    Olivia Burrus, Amanda Curtis, and Laura Herman. 2024. Unmasking AI: Informing Authenticity Decisions by Labeling AI-Generated Content. Interactions31, 4 (2024), 38–42

    2024

  16. [16]

    C2PA. 2024. Introducing Official Content Credentials Icon - C2PA — c2pa.org. https://c2pa.org/post/contentcredentials/. [Accessed 17-01-2024]

    2024

  17. [17]

    Astrid Carolus, Martin J Koch, Samantha Straka, Marc Erich Latoschik, and Carolin Wienrich. 2023. MAILS-Meta AI literacy scale: Development and testing of an AI literacy questionnaire based on well-founded competency models and psychological change-and meta-competencies.Computers in Human Behavior: Artificial Humans1, 2 (2023), 100014

    2023

  18. [18]

    Chaomei Chen. 2010. Information visualization.Wiley Interdisciplinary Reviews: Computational Statistics2, 4 (2010), 387–403

    2010

  19. [19]

    Inyoung Cheong, Alicia Guo, Mina Lee, Zhehui Liao, Kowe Kadoma, Dongyoung Go, Joseph Chee Chang, Peter Henderson, Mor Naaman, and Amy X. Zhang. 2025. Penalizing Transparency? How AI Disclosure and Author Demographics Shape Human and AI Judgments About Writing. arXiv:2507.01418 [cs.CY] https://arxiv.org/abs/2507.01418

    Pith/arXiv arXiv 2025

  20. [20]

    Rune Haubo B Christensen. 2018. Cumulative link models for ordinal regression with the R package ordinal.Submitted in J. Stat. Software35 (2018), 1–46

    2018

  21. [21]

    2013.Statistical power analysis for the behavioral sciences

    Jacob Cohen. 2013.Statistical power analysis for the behavioral sciences. routledge

    2013

  22. [22]

    Marios Constantinides, Himanshu Verma, Shadan Sadeghian, and Abdallah El Ali. 2025. The Future of Work is Blended, Not Hybrid. InProceedings of the 4th Annual Symposium on Human-Computer Interaction for Work (CHIWORK ’25). Association for Computing Machinery, New York, NY, USA, Article 28, 13 pages. https://doi.org/10.1145/3729176.3729202

    arXiv 2025

  23. [24]

    Patrick Corrigan. 2024. LinkedIn Adopts C2PA Standard. https://www.linkedin.com/pulse/linkedin-adopts-c2pa-standard-patrick-corrigan- kwldf/?trackingId=4gnjmapwRsmugUNqwj0fRw%3D%3D Accessed: 2025-01-27

    2024

  24. [25]

    Weiwei Cui, Xiaoyu Zhang, Yun Wang, He Huang, Bei Chen, Lei Fang, Haidong Zhang, Jian-Guan Lou, and Dongmei Zhang. 2019. Text-to-viz: Automatic generation of infographics from proportion-related natural language statements.IEEE transactions on visualization and computer graphics26, 1 (2019), 906–916

    2019

  25. [26]

    Google DeepMind. 2024. SynthID. https://deepmind.google/technologies/synthid/. Accessed: 2024-1-19

    2024

  26. [27]

    Nicholas Diakopoulos, Hannes Cools, Charlotte Li, Natali Helberger, Ernest Kung, Aimee Rinehart, and L Gibbs. 2024. Generative AI in journalism: the evolution of newswork and ethics in a generative information ecosystem

    2024

  27. [28]

    Abdallah El Ali, Karthikeya Puttur Venkatraj, Sophie Morosoli, Laurens Naudts, Natali Helberger, and Pablo Cesar. 2024. Transparent AI Disclosure Obligations: Who, What, When, Where, Why, How. InExtended Abstracts of the CHI Conference on Human Factors in Computing Systems(Honolulu, HI, USA)(CHI EA ’24). Association for Computing Machinery, New York, NY, ...

    arXiv 2024

  28. [29]

    Large Whatever Models

    Passant Elagroudy, Jie Li, Kaisa Väänänen, Paul Lukowicz, Hiroshi Ishii, Wendy E. Mackay, Elizabeth F Churchill, Anicia Peters, Antti Oulasvirta, Rui Prada, Alexandra Diening, Giulia Barbareschi, Agnes Gruenerbl, Midori Kawaguchi, Abdallah El Ali, Fiona Draxler, Robin Welsch, and Albrecht Schmidt. 2024. Transforming HCI Research Cycles using Generative AI...

    arXiv 2024

  29. [30]

    Louis Engelbrecht, Adele Botha, and Ronell Alberts. 2015. Designing the visualization of information.International Journal of Image and Graphics 15, 02 (2015), 1540005

    2015

  30. [31]

    Ziv Epstein, Mengying C Fang, Antonio A Arechar, and David G Rand. 2023. What label should be applied to content produced by generative AI? https://doi.org/10.31234/osf.io/v4mfz

    work page doi:10.31234/osf.io/v4mfz 2023

  31. [32]

    Ziv Epstein, Aaron Hertzmann, Investigators of Human Creativity, Memo Akten, Hany Farid, Jessica Fjeld, Morgan R Frank, Matthew Groh, Laura Herman, Neil Leach, et al. 2023. Art and the science of generative AI.Science380, 6650 (2023), 1110–1111

    2023

  32. [33]

    European Parliament and Council. 2024. Regulation (EU) 2024/1689 of the European Parliament and of the Council laying down harmonised rules on artificial intelligence and amending certain Union legislative acts. https://artificialintelligenceact.eu/article/50/ Official Journal of the European Union

    2024

  33. [34]

    Richard Fletcher and R Nielsen. 2024. What does the public in six countries think of generative AI in news?URL https://reutersinstitute.politics.ox.ac.uk/sites/default/files/2024-05/Fletcher_and_Nielsen_Generative_AI_and_News_Audiences.pdf(2024), 42 pages

    2024

  34. [35]

    Tania Forja-Pena, Berta García-Orosa, and Xosé López-García. 2024. The Ethical Revolution: Challenges and Reflections in the Face of the Integration of Artificial Intelligence in Digital Journalism.Communication & Society37, 3 (Jun. 2024), 237–254. https://doi.org/10.15581/003.37.3.237-254

    work page doi:10.15581/003.37.3.237-254 2024

  35. [36]

    Paul Formosa, Sarah Bankins, Rita Matulionyte, and Omid Ghasemi. 2025. Can ChatGPT be an author?: Generative AI creative writing assistance and perceptions of authorship, creatorship, responsibility, and disclosure.AI and Society40, 5 (June 2025), 3405–3417. https://doi.org/10.1007/s00146- 024-02081-0

    doi:10.1007/s00146- 2025

  36. [37]

    George Fragiadakis, Christos Diou, George Kousiouris, and Mara Nikolaidou. 2025. Evaluating Human-AI Collaboration: A Review and Method- ological Framework. arXiv:2407.19098 [cs.HC] https://arxiv.org/abs/2407.19098

    Pith/arXiv arXiv 2025

  37. [38]

    Dilrukshi Gamage, Dilki Sewwandi, Min Zhang, and Arosha K Bandara. 2025. Labeling Synthetic Content: User Perceptions of Label Designs for AI-Generated Content on Social Media. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems (CHI ’25). Association for Computing Machinery, New York, NY, USA, Article 814, 29 pages. https://doi...

    arXiv 2025

  38. [39]

    Fabrizio Gilardi, Sabrina Di Lorenzo, Juri Ezzaini, Beryl Santa, Benjamin Streiff, Eric Zurfluh, and Emma Hoes. 2024. Willingness to Read AI-Generated News Is Not Driven by Their Perceived Quality.arXiv e-prints, Article arXiv:2409.03500 (Sept. 2024), arXiv:2409.03500 pages. https://doi.org/10.48550/arXiv.2409.03500 arXiv:2409.03500 [cs.CY]

    Pith/arXiv arXiv doi:10.48550/arxiv.2409.03500 2024

  39. [40]

    Gregory Gondwe. 2023. Exploring the multifaceted nature of generative AI in journalism studies: A typology of scholarly definitions.A vailable at SSRN 4465446(2023)

    2023

  40. [41]

    Google. [n.d.]. Disclosing use of altered or synthetic content. https://support.google.com/youtube/answer/14328491?hl=en&co=GENIE.Platform% 3DAndroid Accessed: 2025-01-27

    arXiv 2025

  41. [42]

    Gray, Cristiana Santos, Nataliia Bielova, Michael Toth, and Damian Clifford

    Colin M. Gray, Cristiana Santos, Nataliia Bielova, Michael Toth, and Damian Clifford. 2021. Dark Patterns and the Legal Requirements of Consent Banners: An Interaction Criticism Perspective. InProceedings of the 2021 CHI Conference on Human Factors in Computing Systems(Yokohama, Japan)(CHI ’21). Association for Computing Machinery, New York, NY, USA, Arti...

    arXiv 2021

  42. [43]

    Matthew Groh, Aruna Sankaranarayanan, Nikhil Singh, Dong Young Kim, Andrew Lippman, and Rosalind Picard. 2023. Human Detection of Political Speech Deepfakes across Transcripts, Audio, and Video. arXiv:2202.12883 [cs.HC]

    Pith/arXiv arXiv 2023

  43. [44]

    Lars Guenther, Jessica Kunert, and Bernhard Goodwin. 2025. My New Colleague, ChatGPT? How German Science Journalists Per- ceive and Use (Generative) Artificial Intelligence.Journalism Practice0, 0 (2025), 1–18. https://doi.org/10.1080/17512786.2025.2502794 arXiv:https://doi.org/10.1080/17512786.2025.2502794

    arXiv 2025

  44. [45]

    Okay, whatever

    Hana Habib, Megan Li, Ellie Young, and Lorrie Cranor. 2022. “Okay, whatever”: An Evaluation of Cookie Consent Interfaces. InProceedings of the 2022 CHI Conference on Human Factors in Computing Systems(New Orleans, LA, USA)(CHI ’22). Association for Computing Machinery, New York, NY, USA, Article 621, 27 pages. https://doi.org/10.1145/3491102.3501985

    arXiv 2022

  45. [46]

    Nicolai Brodersen Hansen, Christian Dindler, Kim Halskov, Ole Sejer Iversen, Claus Bossen, Ditte Amund Basballe, and Ben Schouten. 2020. How Participatory Design Works: Mechanisms and Effects. InProceedings of the 31st Australian Conference on Human-Computer-Interaction(Fremantle, WA, Australia)(OzCHI ’19). Association for Computing Machinery, New York, N...

    arXiv 2020

  46. [47]

    Winston Haynes. 2013. Benjamini–hochberg method. InEncyclopedia of systems biology. Springer, 78–78

    2013

  47. [48]

    Jessica He, Stephanie Houde, and Justin D. Weisz. 2025. Which Contributions Deserve Credit? Perceptions of Attribution in Human-AI Co-Creation. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems (CHI ’25). Association for Computing Machinery, New York, NY, USA, Article 540, 18 pages. https://doi.org/10.1145/3706598.3713522

    arXiv 2025

  48. [49]

    Roy S Hessels, Chantal Kemner, Carlijn van den Boomen, and Ignace TC Hooge. 2016. The area-of-interest problem in eyetracking research: A noise-robust solution for face and sparse stimuli.Behavior research methods48, 4 (2016), 1694–1712

    2016

  49. [50]

    Kenneth Holmqvist, Marcus Nyström, and Fiona Mulvey. 2012. Eye tracker data quality: what it is and how to measure it. InProceedings of the Symposium on Eye Tracking Research and Applications(Santa Barbara, California)(ETRA ’12). Association for Computing Machinery, New York, NY, USA, 45–52. https://doi.org/10.1145/2168556.2168563

    arXiv 2012

  50. [51]

    Steffen Holter and Mennatallah El-Assady. 2024. Deconstructing Human-AI Collaboration: Agency, Interaction, and Adaptation.Computer Graphics Forum43, 3 (2024), e15107. https://doi.org/10.1111/cgf.15107 arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1111/cgf.15107

    doi:10.1111/cgf.15107 2024

  51. [52]

    Ignace TC Hooge, Gijs A Holleman, Nina C Haukes, and Roy S Hessels. 2019. Gaze tracking accuracy in humans: One eye is sometimes better than two.Behavior Research Methods51, 6 (2019), 2712–2721

    2019

  52. [53]

    Runsheng Huang, Liam Dugan, Yue Yang, and Chris Callison-Burch. 2024. MiRAGeNews: Multimodal Realistic AI-Generated News Detection. arXiv:2410.09045 [cs.CV] https://arxiv.org/abs/2410.09045

    Pith/arXiv arXiv 2024

  53. [54]

    2024.Information Visualization

    Christophe Hurter, Alexandru Telea, and Bernice Rogowitz. 2024.Information Visualization. CRC Press, 223–262. https://doi.org/10.1201/ 9781003495147-7 Publisher Copyright:©2025 selection and editorial matter, Constantine Stephanidis and Gavriel Salvendy

    2024

  54. [55]

    Vera Liao, Su Lin Blodgett, Alexandra Olteanu, and Adam Trischler

    Angel Hsing-Chi Hwang, Q. Vera Liao, Su Lin Blodgett, Alexandra Olteanu, and Adam Trischler. 2025. ’It was 80% me, 20% AI’: Seeking Authenticity in Co-Writing with Large Language Models.Proc. ACM Hum.-Comput. Interact.9, 2, Article CSCW122 (May 2025), 41 pages. https://doi.org/10.1145/3711020

    doi:10.1145/3711020 2025

  55. [56]

    Ayae Ide, Tory Park, Jaron Mink, and Tanusree Sharma. 2025. Signals of Provenance: Practices & Challenges of Navigating Indicators in AI-Generated Media for Sighted and Blind Individuals. arXiv:2505.16057 [cs.HC] https://arxiv.org/abs/2505.16057

    Pith/arXiv arXiv 2025

  56. [57]

    Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Ye Jin Bang, Andrea Madotto, and Pascale Fung. 2023. Survey of hallucination in natural language generation.ACM computing surveys55, 12 (2023), 1–38

    2023

  57. [58]

    Haiyan Jia, Alyssa Appelman, Mu Wu, and Steve Bien-Aimé. 2024. News bylines and perceived AI authorship: Effects on source and message credibility.Computers in Human Behavior: Artificial Humans2, 2 (2024), 100093. https://doi.org/10.1016/j.chbah.2024.100093

    arXiv 2024

  58. [59]

    Tingting Jiang, Zhumo Sun, Shiting Fu, and Yan Lv. 2024. Human-AI interaction research agenda: A user-centered perspective.Data and Information Management8, 4 (2024), 100078. https://doi.org/10.1016/j.dim.2024.100078

    arXiv 2024

  59. [60]

    Jones and Benjamin K

    Cameron R. Jones and Benjamin K. Bergen. 2025. Large Language Models Pass the Turing Test. arXiv:2503.23674 [cs.CL] https://arxiv.org/abs/ 2503.23674

    Pith/arXiv arXiv 2025

  60. [61]

    Mitt Nowshade Kabir. 2024. Unleashing Human Potential: A Framework for Augmenting Co-Creation with Generative AI. InProceedings of the International Conference on AI Research. Academic Conferences and publishing limited

    2024

  61. [62]

    2016.Using Generalized Linear (Mixed) Models in HCI

    Maurits Kaptein. 2016.Using Generalized Linear (Mixed) Models in HCI. Springer International Publishing, Cham, 251–274. https://doi.org/10.1007/ 978-3-319-26633-6_11 30 Visualizing Human-AI Collaboration Disclosures Woodstock ’18, June 03–05, 2018, Woodstock, NY

    2016

  62. [63]

    John Kirchenbauer, Jonas Geiping, Yuxin Wen, Jonathan Katz, Ian Miers, and Tom Goldstein. 2023. A watermark for large language models. In International Conference on Machine Learning. PMLR, 17061–17084

    2023

  63. [64]

    Pallav Laskar. 2025. Cryptographic Provenance and the Future of Media Authenticity: Technical Standards and Ethical Frameworks for Generative Content.Journal of Computer Science and Technology Studies7, 6 (Jun. 2025), 967–972. https://doi.org/10.32996/jcsts.2025.7.114

    work page doi:10.32996/jcsts.2025.7.114 2025

  64. [65]

    Fan Li, Ya Yang, et al. 2024. Impact of Artificial Intelligence–Generated Content Labels On Perceived Accuracy, Message Credibility, and Sharing Intentions for Misinformation: Web-Based, Randomized, Controlled Experiment.JMIR Formative Research8, 1 (2024), e60024

    2024

  65. [66]

    Jie Li, Hancheng Cao, Laura Lin, Youyang Hou, Ruihao Zhu, and Abdallah El Ali. 2024. User Experience Design Professionals’ Perceptions of Generative Artificial Intelligence. InProceedings of the 2024 CHI Conference on Human Factors in Computing Systems(Honolulu, HI, USA)(CHI ’24). Association for Computing Machinery, New York, NY, USA, Article 381, 18 pag...

    arXiv 2024

  66. [67]

    Jian Li, Jinsong Huang, Jiaxiang Liu, and Tianqi Zheng. 2022. Human-AI cooperation: modes and their effects on attitudes.Telematics and Informatics73 (2022), 101862

    2022

  67. [68]

    Bingjie Liu and Lewen Wei. 2019. Machine authorship in situ: Effect of news organization and news genre on news credibility.Digital journalism7, 5 (2019), 635–657

    2019

  68. [69]

    Jacob A Long, Tabitha Oyewole, Maryam Goli, Jacqueline M Keisler, Saud Alyaqout, Michael D Rodgers, and Arielle N’Diaye. [n.d.]. The Disclosure Dilemma: How AI Attribution Affects Reactions to Public Health Messages. ([n. d.])

  69. [70]

    Chiara Longoni, Andrey Fradkin, Luca Cian, and Gordon Pennycook. 2022. News from Generative Artificial Intelligence Is Believed Less. In Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency(Seoul, Republic of Korea)(FAccT ’22). Association for Computing Machinery, New York, NY, USA, 97–106. https://doi.org/10.1145/3531146.3533077

    arXiv 2022

  70. [71]

    Tambiama Madiega. 2023. Generative AI and watermarking.European Parliament(2023). https://www.europarl.europa.eu/RegData/etudes/BRIE/ 2023/757583/EPRS_BRI(2023)757583_EN.pdf

    2023

  71. [72]

    Andrew Vande Moere and Helen Purchase. 2011. On the role of design in information visualization.Information Visualization10, 4 (2011), 356–371

    2011

  72. [73]

    Moreno-Sánchez, Javier Del Ser, Mark van Gils, and Jussi Hernesniemi

    Pedro A. Moreno-Sánchez, Javier Del Ser, Mark van Gils, and Jussi Hernesniemi. 2026. A design framework for operationalizing trustworthy artificial intelligence in healthcare: Requirements, tradeoffs and challenges for its clinical adoption.Information Fusion127 (2026), 103812. https://doi.org/10.1016/j.inffus.2025.103812

    arXiv 2026

  73. [74]

    Sophie Morosoli, Laurens Naudts, Hannes Cools, Karthikeya Venkatraj, Natali Helberger, and Claes de Vreese. 2025. Public accountability and regulatory expectations for AI in journalism: qualitative evidence from focus groups with Dutch citizens.AI & SOCIETY(2025), 1–13

    2025

  74. [75]

    2005.Ten usability heuristics

    Jakob Nielsen. 2005.Ten usability heuristics. http://www. nngroup. com/articles/ten-usability-heuristics/. 01163

    2005

  75. [76]

    Sachita Nishal and Nicholas Diakopoulos. 2024. Envisioning the Applications and Implications of Generative AI for News Media. arXiv:2402.18835 [cs.CY] https://arxiv.org/abs/2402.18835

    Pith/arXiv arXiv 2024

  76. [77]

    Midas Nouwens, Ilaria Liccardi, Michael Veale, David Karger, and Lalana Kagal. 2020. Dark patterns after the GDPR: Scraping consent pop-ups and demonstrating their influence. InProceedings of the 2020 CHI conference on human factors in computing systems. 1–13

    2020

  77. [78]

    Johannes Oberpriller, Melina de Souza Leite, and Maximilian Pichler. 2022. Fixed or random? On the reliability of mixed-effects mod- els for a small number of levels in grouping variables.Ecology and Evolution12, 7 (2022), e9062. https://doi.org/10.1002/ece3.9062 arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1002/ece3.9062 e9062 ECE-2022-01-00007.R1

    work page doi:10.1002/ece3.9062 2022

  78. [79]

    Anneli Olsen. 2012. The Tobii I-VT fixation filter.Tobii Technology21, 4-19 (2012), 5

    2012

  79. [80]

    Roy Perlis. 2025. AI Disclosure and Patient Consent in Health Care.JAMA334, 11 (09 2025), 961–961. https://doi.org/10.1001/jama.2025.14026 arXiv:https://jamanetwork.com/journals/jama/articlepdf/2837952/jama_perlis_2025_en_250008_1757361763.57499.pdf

    arXiv 2025

  80. [81]

    Stanislaw Piasecki, Sophie Morosoli, Natali Helberger, and Laurens Naudts. 2024. AI-generated journalism: Do the transparency provisions in the AI Act give news readers what they hope for?Internet Policy Review13, 4 (2024), 1–28. https://doi.org/10.14763/2024.4.1810

    work page doi:10.14763/2024.4.1810 2024

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 3, 2026.