Credit: Matheus Bertelli from Pexels

Science is entering a massive publishing boom, in large part due to artificial intelligence. New research published in the journal Science has revealed that scientists who use large language models (LLMs) like ChatGPT are producing significantly more papers across many fields. The technology is also helping to level the playing field for researchers whose first language isn’t English.

The growing use of AI in scientific research has sparked concerns about shoddy work and machines making things up. But this new analysis also reveals that papers produced with LLMs use more complex language and cite a wider array of sources.

AI analysis

To quantify the effect of AI on scientific publishing, researchers from Cornell University and the University of California, Berkeley analyzed nearly 2.1 million study abstracts published on three major preprint servers between January 2018 and June 2024. These papers were posted before they underwent peer review.

To perform their analysis, the team used chatbot GPT-3.5 Turbo-0125 to generate AI-written versions of abstracts published before 2023. Then they checked the AI writing for patterns that distinguish it from human text. With this knowledge, they created an algorithm to scan newer papers for similar patterns to flag AI-assisted work. They also tracked authors over time so they could measure how their publication volume changed.

Productivity surge

They found that when scientists use AI, their productivity soared. The biggest jump was in the social sciences and humanities, where output increased by 59.8%, while biology and life sciences saw a 52.9% increase. Meanwhile, in physics and math, the scientists report a 36.2% boost. "LLM adoption is associated with a large increase in researchers’ scientific output," wrote the team.

One of the most fascinating findings was the massive increase in productivity from non-English-speaking countries. Most top journals require manuscripts to be written in high-level English, which has long disadvantaged these scientists. But with AI handling some of the workload, researchers from Asia saw their output jump by as much as 89% in some cases.

However, the study authors also issued a warning regarding AI and quality. While the machines can make papers sound more professional, this can be a trap. Historically, sophisticated writing was a sign of high-quality research, but now the opposite can sometimes be true. The study found that the more complex the AI-generated writing was, the less likely the paper was to be high quality. In other words, good writing can mask weak ideas.

The clear message from the study authors is that we can no longer judge a paper by how smart the language appears. "As traditional heuristics break down, editors and reviewers may increasingly rely on status markers such as author pedigree and institutional affiliation as signals of quality, ironically counteracting the democratizing effects of LLMs on scientific production."

To safeguard scientific integrity, the researchers propose several measures, including that institutions implement deeper checks and even specialized "AI-based reviewer agents" to help distinguish between human writing and machine-generated writing.

Written for you by our author Paul Arnold, edited by Gaby Clark, and fact-checked and reviewed by Robert Egan—this article is the result of careful human work. We rely on readers like you to keep independent science journalism alive. If this reporting matters to you, please consider a donation (especially monthly). You’ll get an ad-free account as a thank-you.

More information: Keigo Kusumegi et al, Scientific production in the era of large language models, Science (2025). DOI: 10.1126/science.adw3000

© 2025 Science X Network

Citation: Scientists who use AI tools are publishing more papers than ever before (2025, December 21) retrieved 21 December 2025 from https://phys.org/news/2025-12-scientists-ai-tools-publishing-papers.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.