AI evidence extraction

AI-Enhanced Scholarly Communication: Transforming Peer Review, Knowledge Dissemination, and Academic Publishing Workflows in the Digital Era

M. A. Mohamed Salih, Tahir Gul, Syeda Sumblah Bukhari, Anam Majeed · International Journal of Ethical AI Application · 2025

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

9/10

Relevance

0/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.64229/4qp06e84

Methodology & findings

Study design

Qualitative multiple-case methodology combining semi-structured interviews (n=20 participants across four stakeholder groups: journal editors, peer reviewers, academic authors, and scholarly publishing technologists/admins) and document analysis (publisher policy documents, AI tool manuals, pilot-program reports, and editorial guidelines) with thematic coding following Braun and Clarke's six-phase approach..

Main result

The study found that "AI tools significantly reduce time for administrative checks (e.g. formatting, citations, plagiarism)" with turnaround time decreased by a maximum of 40%. Additionally, "the human reviewers valued computer-generated intelligence—the potential to identify conflicts of interest, statistical or citation anomalies" and "It does not criticize the hypothesis, but it shows strange trends that I would otherwise not see," demonstrating that AI serves effectively as an assisted decision-aid rather than a replacement for human judgment.

Research paradigm

Qualitative interpretivism

Author conclusions

The authors conclude that "This article brings out the revolution AI has on educational communication. The efficiency can increase dramatically with AI tools because it is assumed that the routine checks could be fully automated and the process of reviews could occur more frequently with more metadata created. They are very powerful aides to human judgment and can offer information-based arguments to its reviewers without augmenting the critical thinking. However, the problems of opacity, bias, and homogenizing pressure of extrinsic demands toward different authors cannot be dismissed as groundless and they deserve to be addressed." The authors recommend adoption of "a hybrid human, and AI model: the human raters are to be supplemented, but not replaced by the AI."

Risk of bias

Selection bias: Purposive sampling may not represent all stakeholder perspectives; participants selected for 'firsthand experience' may be early adopters or more favorable to AI; Interviewer bias: No mention of inter-rater reliability or blinding in the thematic coding process; Attrition: Not reported; Temporal bias: Snapshot of current practices; future adoption patterns unknown; Geographic bias: No indication of international representation in the 20 participants; Confirmation bias: No explicit discussion of how researchers handled contradictory findings; Selection bias: Purposive sampling may not represent all stakeholder perspectives equally; Participation bias: Self-selected interview participants may have stronger views on AI; Researcher bias: Single qualitative analysis without mention of inter-rater reliability; Geographic bias: Likely overrepresentation of Western, English-speaking institutions in study population; Selection bias: purposive sampling of participants with firsthand AI experience may not represent broader academic community perspectives; Self-selection bias: participants willing to discuss AI tools may differ from reluctant adopters; Social desirability bias: participants in interviews may express views aligned with perceived institutional expectations; Limited geographic/disciplinary representation: sampling strategy not detailed across institution types

Open questions raised

The authors identify the need for future quantitative evaluation including measurement of review turnaround time indicators, reviewer satisfaction levels, adoption rates between disciplines, and inequalities in acceptance rates between different author groups. They recommend longitudinal research to track how AI influences academic standards over time and whether hybrid models develop into new forms of human-machine collaboration.
Future research should include: quantitative evaluation of changes in review turnaround time, reviewer satisfaction levels, adoption rates between disciplines, and inequalities in acceptance between different author groups. Longitudinal research is needed to track how AI influences academic standards over time and whether hybrid models develop into new forms of human-machine collaboration.
The authors identify future research directions: "To further evaluate the effect of the AI integration, additional research may include quantitative evaluation of the changes in the indicators of review turnaround time, the level of satisfaction of the reviewers, the rate of adoptions between disciplines, and any inequalities in acceptance between different groups of authors. Longitudinal research would be possible to track the development of how AI influences academic standards and whether hybrid models are developing into new styles of human machine rule."

Extracted from: pdfAgreement 80%

Explore related topics

Scholarly Publishing Scholarly Infrastructure Literature Review