Ownership in AI-Assisted Everyday Tasks
Megan Wei, Melanie Subbiah, Audrey Lee, Annya Dahmani, Dave Edwards, Helen Edwards, Ellie Pavlick
cs.AI · cs.HC
Quick verdict
Exploratory survey (n=64): AI-task ownership hinges on process, voice, comprehension; insightful but biased sample and untested percentages limit claims.
Summary
The paper reports an exploratory qualitative survey asking when AI-assisted work still feels like one's own. Of 117 people who began, 52 completed the survey and 64 gave at least one task reflection; 49 described both a high- and a low-ownership task. Participants were recruited via professional networks and social media, August 6-27, 2026, and self-selected two recent AI-assisted tasks to answer Likert and free-text questions. The themes: ownership hinges on process — leading, iterating, rewriting — rather than merely approving AI output; a distinctive personal voice and model personalization sustain it; people report high ownership over tasks they could not do without AI when they own the vision, such as illustrating children's books; incomprehensible outputs erode ownership; and willingness to disclose AI use is decoupled from pride, driven instead by community norms and fear of credit erasure. Descriptively, 42% of high-ownership task respondents said they could definitely do the task without AI, versus 63% for low-ownership tasks. Key limitations: a self-selected, heavily biased sample (79.7% white, 71.7% US-based, heavy AI users), about 55% attrition, no inferential statistics, and coding by fixed keyword rules and LLM judging without reported reliability checks.
Review complete
Assessment at a glance
4
risk flags identified
● high 0
● medium 3
● low 1
4 strengths · 5 weaknesses · 5 author questions
AI text signal = Binoculars zero-shot detector score over 20 sampled passages (Fast-DetectGPT co-reported) — absence does not prove human authorship, and a flag does not imply misconduct.
Levels reflect verification complexity — not a quality score. Evolve never rates your paper.
Key claims under review
- Felt ownership of AI-assisted work depends primarily on the collaboration process — leading, iterating, rewriting — rather than the output alone.
- People can report high ownership over tasks they could not have completed without AI when they own the project's vision (42% vs 63% 'definitely yes' able-without-AI, Figure 1).
- Loss of personal voice and inability to comprehend or evaluate outputs erode felt ownership.
- Willingness to disclose AI use is decoupled from pride or ownership and is instead driven by community norms and fear of credit erasure.
Strengths
Contrastive within-subject design
Each participant described one high- and one low-ownership AI task in randomized order (Section 2, Appendix A.3); 49 did both, enabling same-person mechanism contrasts exemplified in Table 5's 13 paired anecdotes (e.g., the artisan steering a car diagnosis vs passively receiving the answer).
Unusual transparency about limits and materials
Footnote 1 and Appendix C openly flag the biased, heavy-user sample; the full Qualtrics instrument is reproduced in Appendix A; results are posted publicly (aiownership.github.io); per-figure sample sizes are reported throughout (Table 1, Figures 4-6).
A counterintuitive, actionable finding
Strong theoretical anchoring
Weaknesses
Selection circularity
Descriptive-only quantitative claims
Unvalidated automated coding
Attrition and shifting sample sizes
Biased, affluent sample beneath universal phrasing
Full Review Report
$1.99 · one-timeUnlock the complete preliminary analysis Evolve produced for this paper.
Report preview
Scope & Claims
arXiv:2609.20658 (17 Sep 2026; Brown, CMU, UC Berkeley, Artificiality Institute) asks a well-being question rather than a capability one: when does work completed with an LLM still feel like ours? The paper is explicitly exploratory (an 'informal survey,' footnote 1) and advances five linked claims: (1) ownership tracks collaboration process — leading, iterating, rewriting — rather than output; (2) a distinctive personal voice and model personalization sustain ownership; (3) people can feel high ownership over tasks they could not have done unaided, provided they own the vision; (4) comprehension of output is a near-precondition; (5) willingness to disclose AI use is decoupled from pride, governed instead by community norms and fear of credit erasure. Four research agendas (process visibility, personalization, enabling-the-impossible, honest disclosure) close Section 4.
The remaining sections cover Methodology Assessment, Statistical Evidence, Results & Reproducibility, Limitations & Risks, and Verdict & Recommendations.
- All 4 strengths & 5 weaknesses, analyzed in detail
- 4 risk flags with severity, section & reviewer questions
- Methods & statistical evidence assessment
- 5 questions a peer reviewer would ask
- Full six-dimension evaluation report & recommendations
One-time purchase · this paper only
One-time purchase, this paper only. AI-generated preliminary analysis — not peer review.
Frequently asked questions
What is the main finding of 'Ownership in AI-Assisted Everyday Tasks'?
The survey finds that people feel ownership over AI-assisted work when they lead the process — iterating, rewriting, personalizing the model — and disown work where they merely approved AI output. Ownership also survives when people own the vision for a task they could not execute without AI, and disclosure of AI use is driven more by community norms than by pride in the work.
Who was surveyed and how many participated?
117 people started the August 2026 online survey; 52 completed it and 64 described at least one AI-assisted task (49 described both a high- and a low-ownership task). The sample skewed heavily toward white, US-based, highly educated, frequent AI users recruited through the authors' professional networks and social media.
Did participants feel ownership over work they could not have done without AI?
Yes, counterintuitively: only 42% of high-ownership task respondents said they could definitely have done their task without AI, versus 63% for low-ownership tasks. Owning the vision was enough — for example, writers who could not illustrate still felt strong ownership over AI-illustrated children's books, comparing AI to a hired illustrator.
Is the paper's data public?
The full survey instrument is reproduced in the paper's Appendix A, and the authors state that survey results are viewable at aiownership.github.io. No analysis code or reliability validation for the automated codings is provided in the paper itself.
About this review
This is an AI-generated preliminary review by Evolve.Pub, produced from the openly available text of the paper. It is not peer review and does not represent the authors' views. The canonical record of the paper is at the publisher link above.
- Figures arrive as ASCII number dumps; bar-to-value mapping was inferred from legends (e.g., Figure 5's 74%/32% pride values come from the label sequences), not from rendered charts.
- The stated public results link (aiownership.github.io) was not visited; data completeness and quality are unverified.
- Free-text reflections are available only via selected quotes; the underlying response corpus cannot be assessed for quote representativeness.
Go deeper on this paper
Wrote a paper? Stress-test it before reviewers do.
Upload your manuscript for the same structured scrutiny — reasoning, statistics, evidence, figures, and reproducibility.