AI evidence extraction

Governing Artificial Intelligence in the Evolving Academic Publishing Ecosystem

Enis Karaarslan · Journal of Metaverse · 2026

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

9/10

Relevance

2/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.57019/jmv.1899774

Methodology & findings

Study design

Narrative policy analysis and literature review synthesizing current AI applications in editorial workflows, theoretical frameworks from institutional theory and algorithmic accountability research, and qualitative risk-benefit assessment.

Main result

The paper finds that "the integration of AI into editorial workflows can be interpreted through established theoretical perspectives on technological and institutional change" and that "the integration of AI tools into editorial workflows is no longer a question of whether it will occur, but rather how it should be governed." The authors conclude that while "AI undeniably accelerates desk-screening in high-volume submission environments, supports non-native authors, and enhances the detection of image manipulation or plagiarism," systemic vulnerabilities including "LLM hallucinations can severely distort scholarly claims" and "VLMs trained on biased datasets might misclassify figures originating from underrepresented domains" must be carefully managed through governance frameworks.

Reports effect sizes.

Research paradigm

Interpretive/argumentative framework grounded in institutional theory and technology governance

Author conclusions

The authors conclude that "The intersection of AI with editorial workflows presents both significant operational opportunities and inherent systemic risks for scholarly journals. Our objective is to leverage the efficiencies and analytical capabilities of AI while safeguarding against vulnerabilities such as confidentiality breaches, algorithmic bias, and unverified content generation. The proposed policy framework establishes a sustainable balance by permitting transparent AI assistance while explicitly preserving human editorial judgment and academic equity."

Risk of bias

Author bias as journal leadership (Journal of Metaverse editorial board); potential publication bias in cited literature; lack of systematic evidence synthesis may overweight certain risk narratives; no explicit methodology for selecting cited studies; potential confirmation bias in framing AI risks and benefits.; Potential institutional bias as the authors are editors of the Journal of Metaverse; Limited empirical validation of policy recommendations; Absence of systematic evidence synthesis across publisher policies; Risk of publication bias in cited studies regarding AI benefits; Selection bias in cited literature - primarily draws from major publishers' perspectives; Institutional perspective bias - editorial leadership viewpoint may not represent author perspectives; Temporal bias - rapid evolution of AI tools means some cited evidence may quickly become outdated; Geographic bias - policies described are primarily from major publishers in developed nations

Limitations

The paper acknowledges that "although the accuracy and effectiveness of AI detection tools are currently questioned," and notes that "several publishers and editorial platforms have begun to operationalize these concerns through measurable indicators" but recognizes limitations in operationalization
The authors also state "while such metrics should never replace human editorial judgment, they can serve as practical monitoring tools" rather than definitive measures
Additionally, the paper notes that "Hallucination can be reduced with systems like RAG, but cannot be avoided," indicating inherent limitations of current technology.

Open questions raised

The authors identify several gaps and future research directions: (1) the need for validated and reliable AI detection tools for undisclosed AI usage; (2) ongoing evolution of AI-agent workflows in editorial systems requiring continuous policy adaptation; (3) need for empirical monitoring of whether AI-driven workflow changes affect overall quality of published work; (4) specific methodological challenges in metaverse research involving synthetic environments and AI-generated avatars requiring further investigation; (5) the need to distinguish authentic participant activity from automated or synthetic agents in metaverse experiments.
Need for more validated AI detection tools for identifying AI-assisted writing
Lack of empirical evidence on the long-term impact of AI-driven submission surges on journal quality metrics
Limited understanding of equity implications of AI tool access across global institutions
Insufficient research on how AI affects quality of peer review in practice
Need for domain-adaptive evaluation strategies for VLM-based synthetic image forensics in publishing contexts

Extracted from: pdfAgreement 67%

Explore related topics

Institutional Policy AI Governance Position Paper