Original aggregate data
AI-Assisted Manuscript Review Quality Snapshot 2026
Published: 2026-07-18 · Page updated: 2026-08-09 · Version 2026-07-18-v1
Key finding
In snapshot 2026-07-18-v1, the average LLM-as-judge quality score for completed AI-assisted manuscript reviews was 7.21/10 (n=68). Overall scores use a 0–10 scale; dimension scores use 0–5. This measures generated-review quality, not manuscript acceptance probability.
Score bands in this snapshot: 0–6 = 10 (15%); >6–8 = 50 (74%); >8–10 = 8 (12%). Percentages are derived from the published counts only.
Sample size
n=68
Average overall
7.21/10
Strong band (>8–10)
8
Period end
2026-07-17
Score distribution
| Band (overall 0–10) | n | % |
|---|---|---|
| 0–6 | 10 | 15 |
| >6–8 | 50 | 74 |
| >8–10 | 8 | 12 |
Quality dimensions (0–5)
Dimension means are on a 0–5 scale (not /10). Values below come only from this versioned snapshot.
Citation accuracy
3.88/5
n=68
Specificity
4.66/5
n=68
Comprehensiveness
3.39/5
n=68
Actionability
4.43/5
n=68
Tone
5/5
n=68
Methodology and privacy
Operational quality assurance: completed review jobs are scored by a separate four-panel LLM-as-judge. We publish only anonymized aggregates (overall mean, score bands, dimension means). No manuscript text, title, author, institution, DOI, email, file name, or user identifier is exported. Groups smaller than the minimum group size are suppressed. The JSON snapshot is versioned and does not update live.
Minimum group size for publication: n=20. Source snapshot file: data/reports/2026-07-18-v1.json.
Limitations
- This is operational quality-assurance data, not peer-reviewed research.
- Scores evaluate generated review quality, not manuscript acceptance probability.
- The submitted-manuscript population is self-selected.
You may cite this snapshot
If this operational dataset informed your writing, comparison, or tooling discussion, you may cite it as follows (not peer-reviewed research):
Börcek AO. AI-Assisted Manuscript Review Quality Snapshot 2026. Review My Manuscript; 2026. Version 2026-07-18-v1. Available from: https://reviewmymanuscript.com/reports/ai-review-quality-2026
License: CC BY 4.0 — attribution required; not peer-reviewed research.
FAQ
- What does the 7.21/10 score mean?
- It is the mean overallScore from 68 completed quality-judge runs in this snapshot (2026-07-18-v1). It rates the generated review report, not whether a journal would accept the manuscript.
- Can I cite this page?
- Yes. Cite Börcek AO. AI-Assisted Manuscript Review Quality Snapshot 2026. Review My Manuscript; 2026. Version 2026-07-18-v1. URL: https://reviewmymanuscript.com/reports/ai-review-quality-2026. Treat it as operational QA data, not peer-reviewed research.
- Are dimension scores on the same scale as the overall score?
- No. Dimension averages (citation accuracy, specificity, comprehensiveness, actionability, tone) are on a 0–5 scale. The overall score is on a 0–10 scale.