Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 74 of the 74 references it cites were found in a registry or an index; 16 of the 48 sources its claims cite were checked for support.

Which Contributions Deserve Credit? Perceptions of Attribution in Human-AI Co-Creation

Open access
He, Jessica, Houde, Stephanie, Weisz, Justin D. · 2025 · Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems

Abstract

AI systems powered by large language models can act as capable assistants for writing and editing. In these tasks, the AI system acts as a co-creative partner, making novel contributions to an artifact-under-creation alongside its human partner(s). One question that arises in these scenarios is the extent to which AI should be credited for its contributions. We examined knowledge workers’ views of attribution through a survey study (N=155) and found that they assigned different levels of credit across different contribution types, amounts, and initiative. Compared to a human partner, we observed a consistent pattern in which AI was assigned less credit for equivalent contributions. Participants felt that disclosing AI involvement was important and used a variety of criteria to make attribution judgments, including the quality of contributions, personal values, and technology considerations. Our results motivate and inform new approaches for crediting AI contributions to co-created work.

Overview

CoChat Summary

TL;DR

A scenario-based survey of 155 knowledge workers found that people assign different levels of authorship credit to an AI co-creative partner depending on the type, amount, and initiative of its contribution, and consistently assign AI less credit than a human partner for equivalent contributions. The authors argue this granularity shows binary "AI was used / not used" disclosure policies are insufficient and motivate AI-specific attribution frameworks.

Key findings

  • Credit varied by contribution type: content contributions warranted more credit than form contributions (M = −1.03, SD = 1.20 vs. M = −1.94, SD = 0.86; W = 131807.5, p < .001, r = .40, moderate). For AI partners specifically, the same contrast held (W = 30531.5, p < .001, r = .43, moderate). Spelling/grammar correction straddled no-credit and acknowledgment; synthesizing information warranted equal authorship.
  • AI was rated lower than a human partner across nearly all contribution natures, with 7 instances where the AI's contribution warranted a lower categorical level of credit. E.g., narrowing scope merited secondary authorship from a human (M = −1.38, SD = 0.84) but only acknowledgment from AI (M = −1.69, SD = 0.99). Exceptions were fact checking (p = .06), adding new ideas (p = .09), and synthesizing information (not significant).
  • For equal writing amounts, the AI's 95% CI [−0.89, −0.45] fell in the secondary-authorship range while the human's [−0.30, −0.01] fell in the equal-authorship range. When a partner wrote everything, humans were rated sole author (M = 2.70, SD = 0.93) but AI was not (M = 2.05, SD = 1.56; W = 3655.0, p < .001, r = .31, moderate).
  • Proactive AI that wrote complete text received more credit (M = 2.48, SD = 1.31) than AI writing on request (M = 1.85, SD = 1.67; W = 3430.0, p = .002, r = .27, small); no significant difference for proactive vs. requested recommendations. Contribution type was rated most important in both conditions (AI: M = 4.39, SD = 0.93; human: M = 4.46, SD = 0.82).

Methods

  • Scenario-based survey with a 2 × 3 between-subjects design: writing partner (human vs. AI) × writing context (academic, professional, technical). Participants were employees of a large multinational technology company recruited via internal Slack channels; 184 responses were filtered to a final N = 155 (human N = 78, AI N = 77), excluding those with no generative AI experience, those failing an attention check, and those selecting "Unsure" throughout.
  • Participants rated authorship credit for 18 scenarios spanning contribution type, amount, and initiative on a 7-point scale (sole self-author to sole partner-author, plus "Unsure"), converted to numeric scores of [−3, +3] with "Equal" at 0. They also rated the importance of the three dimensions on a 5-point scale and answered open-ended questions.
  • Because Shapiro-Wilk tests showed non-normal, self-credit-skewed distributions, pairwise partner comparisons used Wilcoxon rank-sum tests (reported as W with effect size r); writing context was omitted from analysis after ANOVA showed it was not a significant predictor. Open-ended responses were analyzed with reflexive thematic analysis, yielding 6 themes and 19 subthemes.

Limitations

  • Stated: respondents were employees of an international technology company with prior generative AI experience, so views may not reflect broader populations; institutional restrictions prevented collecting age or gender identity; only three writing contexts and only written work were studied, so results may not extend to other content forms (images, video, music, code); the study examined isolated individual contributions rather than complex aggregated workflows; scenarios were hypothetical, so responses may differ from real-world practice; and perceptions may shift as generative AI evolves.
  • Evident from the text: the authorship credit scale was skewed toward self-credit, which the authors attribute to the second-person framing of the scenarios ("As part of your work, you're writing..."), potentially inflating self-credit across conditions.

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

Integrity signals

Registry

Crossref · 10.1145/3706598.3713522

Cites

74 of 74 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

13 Valid · 0 Not valid · 9 Not checked · 101 Own results and interpretations

53 more claims

Create a free account to see all 123 claims and to verify the 22 references still open.

See all claims — free account
  • Probably not a substantial claim. 47
  • It may state two separate claims as one. 31
  • The quote only partly states this claim. 2

Create a free account to see these claims.

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.

2 of 2

Citations

Cites

Show all 55 references

Research works printed but not linked: 12

  1. ACM. 2023. ACM Policy on Authorship. https://www.acm.org/publications/policies/new-acm-policy-on-authorship.
  2. arXiv. [n. d.]. arXiv moderation. https://info.arxiv.org/help/moderation/index.html#policy-for-authors-use-of-generative-ai-language-tools.
  3. Alex Calderwood, Vivian Qiu, Katy Ilonka Gero, and Lydia B Chilton. 2020. How novelists use generative language models: An exploratory user study.. In HAI-GEN+ user2agent@ IUI.
  4. Creative Commons. [n. d.]. License Chooser. Retrieved 09-Sep-2024 from https://chooser-beta.creativecommons.org
  5. Elsevier. [n. d.]. The use of generative AI and AI-assisted technologies in writing for Elsevier. https://www.elsevier.com/about/policies-and-standards/the-use-of-generative-ai-and-ai-assisted-technologies-in-writing-for-elsevier.
  6. Xiaomeng Hu Pin-Yu Chen and Tsung-Yi Ho. 2023. Radar: Robust ai-text detection via adversarial learning. Advances in Neural Information Processing Systems 36 (2023) 15077–15095.
  7. Michael Muller, Stephanie Houde, Gabriel Gonzalez, Kristina Brimijoin, Steven I Ross, Dario Andres Silva Moran, and Justin D Weisz. 2024. Group Brainstorming with an AI Agent: Creating and Selecting Ideas. In International Conference on Computational Creativity.
  8. Michael Muller and Justin Weisz. 2022. Frameworks for Collaborating Humans and AIs: Sequence and Sociality in Organizational Applications. In CHIWORK.
  9. Springer Nature. [n. d.]. Artificial Intelligence (AI). https://www.nature.com/nature-portfolio/editorial-policies/ai.
  10. Wiley. 2023. Generative Artificial Intelligence Guidelines for Authors. Retrieved 15-Aug-2024 from https://onlinelibrary.wiley.com/pb-assets/assets/15405885/Generative%20AI%20Policy_September%202023-1695231878293.pdf
  11. Daijin Yang, Yanpeng Zhou, Zhiyuan Zhang, Toby Jia-Jun Li, and Ray LC. 2022. AI as an Active Writer: Interaction strategies with generated text in human-AI collaborative fiction writing. In Joint Proceedings of the ACM IUI Workshops , Vol. 10. CEUR-WS Team, 1–11.
  12. Ceyao Zhang Kaijie Yang Siyi Hu Zihao Wang Guanghe Li Yihang Sun Cheng Zhang Zhaowei Zhang Anji Liu Song-Chun Zhu et al. 2023. Proagent: Building proactive cooperative ai with large language models. CoRR (2023).

Cited by

See all 38

Metadata

Authors
He, J., Houde, S., & Weisz, J. D.
Affiliations
  • IBM (United States) (US)
  • IBM Research - Cambridge (United States) (US)
Venue
Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems
Publisher
ACM
Published
2025-04-25
Pages
1-18
OpenAlex
W4409749619
Cited by
36
References
55
Citations by year
2026: 31 · 2025: 5
FWCI
34.14
Type
conference paper
Language
en
Open access
Locations
  • CHI Conference on Human Factors in Computing Systems (publishedVersion, cc-by-nc-nd)
  • arXiv (Cornell University) (submittedVersion)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Law, AI, and Intellectual Property
  • Physical Sciences › Engineering › Mechanical Engineering › Design Education and Practice
  • Social Sciences › Human Factors and Ergonomics › Human-Automation Interaction and Safety
Keywords
human-AI co-creation, attribution, credit allocation, human-AI collaboration, AI authorship

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.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-11-01, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-01 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

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

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Which Contributions Deserve Credit? Perceptions of Attribution in Human-AI Co-Creation. https://app.cochat.ai/papers/10.1145/3706598.3713522
The paper
He, J., Houde, S., & Weisz, J. D. (2025). Which Contributions Deserve Credit? Perceptions of Attribution in Human-AI Co-Creation. Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3706598.3713522

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.