AI evidence extraction

AI, originality, and attribution: Researchers’ perspectives on distinguishing contributions

Yanyi Wu, Xinyu Lu · Accountability in Research · 2025

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

1/4

Quality (LMQS)

Evidence

Citations

3.02

FWCI

Top 10%

Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1080/08989621.2025.2536817

Methodology & findings

Study design

Semi-structured interviews with reflexive thematic analysis informed by Attribution Theory

Sample

N = 18, 5 groups

Primary method

Reflexive thematic analysis informed by Attribution Theory. No quantitative statistical methods reported in abstract.

Main result

Researchers predominantly conceptualize AI as a sophisticated tool requiring significant human direction, rather than a genuine collaborator. "To navigate attributional ambiguity, they rely on subjective heuristics - such as 'gut feelings' of ownership and using the labor of the research process as a proxy for conceptual contribution." This creates significant ethical tensions and a desire for clearer, more nuanced guidelines.

Reports effect sizes.

Research paradigm

Qualitative interpretivist / constructivist epistemology

Author conclusions

Researchers "face cognitive and practical challenges applying traditional integrity norms to AI-assisted work. Findings highlight the need for critical dialogue, reflective practices, and nuanced guidelines to uphold research integrity and thoughtfully integrate human value with machine capabilities."

Risk of bias

Small sample size (n=18) may limit generalizability; Potential selection bias in researcher recruitment (self-selected volunteers); Disciplinary variation (STEM, Social Sciences, Humanities) may introduce confounding; Reflexive thematic analysis subject to researcher interpretation bias; Selection bias: Self-selected researcher sample (18 participants across disciplines); Recall bias: Subjective retrospective reports of research practices; Social desirability bias: Researchers may present ethically favorable accounts of their practices; Disciplinary representation: Only 18 researchers across STEM, Social Sciences, and Humanities may not adequately represent all fields; Small sample size (n=18); Self-selection bias in interview recruitment; Potential social desirability bias in researcher responses about authorship practices; Limited disciplinary representation despite claim of 'diverse disciplines'; No information on interviewer training or inter-rater reliability for thematic analysis

Open questions raised

Gap in understanding how individual researchers subjectively perceive and navigate ambiguities around AI contributions in practice; need for clearer, more nuanced guidelines for AI attribution in research.
The authors identify a gap in understanding how individual researchers subjectively perceive and navigate ambiguities around AI contributions in practice, impacting research integrity. They call for clearer, more nuanced guidelines and critical dialogue on distinguishing human versus AI contributions.
Gap in understanding how individual researchers subjectively perceive and navigate AI-related ambiguities in practice; need for clearer, more nuanced guidelines on AI attribution and authorship

Data: not_statedCode: not_statedExtracted from: pdfAgreement 69%

Explore related topics

Authorship & Attribution Human-AI Collaboration Qualitative