12,445 papers · continuously updated · last export: 10 Aug 2026livingmeta.ai
← Browse all papers
AI evidence extraction

Scholarly Communications in 2025: An Aerial Evaluation of a System Challenged by AI and Much More

Learned Publishing · 2026

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

9/10
Relevance
1/4
Quality (LMQS)
E
Evidence
0
Citations
0.00
FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1002/leap.2056

Methodology & findings

Study design

Qualitative and quantitative analysis of semi-structured interview data from the 2025 Harbingers project.

Sample

N = 62, 1 group

Primary method

Data analyzed by country, age, gender and subject using summarisation, codification and quantification of interview data. Specific statistical software or tests not reported.

Main result

The study found widespread agreement among early career researchers that "citations remain the main reputational currency, that traditional publishing outlets are here to stay, that peer review badly needs fixing and, most importantly, that AI is impacting on a very wide front across the scholarly enterprise." Additionally, there is overarching concern that "the quality of research output is being undermined, through bad actors and AI ghostwriters."

Reports effect sizes.

Research paradigm

Mixed methods (qualitative and quantitative)

Author conclusions

The authors conclude that "AI is impacting on a very wide front across the scholarly enterprise" and that there are pervasive concerns about research quality. They note that while "Findings are based on a relatively small, convenience sample, so they should not be regarded as definitive, rather as pointers," the study provides rare empirical data on ECR perspectives on AI and scholarly communications.

Risk of bias

Convenience sampling bias - non-random selection of participants; Small sample size (n=62) limiting representativeness; Selection bias from self-selection in interview participation; Potential interviewer bias in data collection and coding; Selection bias due to convenience sampling; Sample size limitations (n=62); Potential selection bias favoring ECRs willing to discuss AI and scholarly communications; Geographic and disciplinary representation may not be representative; Selection bias: convenience sampling of 62 ECRs from 6 countries may not be representative; Geographic limitation: only 6 countries represented; Sample size: relatively small sample limits statistical power and generalizability; Potential volunteer bias: self-selected participants may have stronger views on AI and scholarly communications

Limitations

  • The authors explicitly state that "Findings are based on a relatively small, convenience sample, so they should not be regarded as definitive, rather as pointers." Additionally, the study employed a non-random, convenience sampling approach which limits generalizability.

Open questions raised

  • The authors note that extensive literature review shows this is "a rare study of ECRs," suggesting a gap in research specifically focused on early career researcher perspectives on AI and scholarly communications.
  • The authors identify this as "a rare study of ECRs" on AI and scholarly communications, suggesting a significant gap in the literature regarding early career researcher perspectives on these topics.
  • The authors note that "an extensive literature review shows that this is a rare study of ECRs," suggesting a gap in empirical research on early career researchers' perspectives on scholarly communications and AI.
Data: not_statedCode: not_statedExtracted from: pdfAgreement 71%

Explore related topics

Multi-DomainSurvey

Related papers

← Browse all papers