AI Fuels Academic Submissions as Writing Quality Falls

Generative AI has sharply lowered the cost of drafting and polishing academic manuscripts, colliding with a publish-or-perish system that rewards output. The shift matters because peer review still depends largely on unpaid human judgment, leaving journals to process more submissions without a comparable increase in editorial capacity. The evidence points most directly to weaker prose and narrower reviews, rather than proving that the underlying science itself has deteriorated.

An Organization Science task force reported on April 27 that it examined 6,957 initial submissions and 10,389 referee reports from January 2021 through February 2026. Submission volume rose 42% after ChatGPT’s November 2022 debut, and by February 2026 most manuscripts showed some degree of AI-assisted writing. Readability in January 2026 was 1.28 standard deviations below its January 2021 level, while more than 30% of current reviews showed detectable AI use and tended to focus more on theory than data.