Chatbot Fingerprints Found in Thousands of Biomedical Research Papers, Study Reveals
A study finds AI like ChatGPT influences 13.5% of recent biomedical abstracts, marked by overuse of certain words. Usage varies by country and journal selectivity.

Tracking Chatbot Influence in Biomedical Research Papers
Scientists have identified clear signs that chatbots like ChatGPT are increasingly contributing to the writing of biomedical research papers. A recent study published in Science Advances reveals how the use of certain words in paper abstracts has surged since the release of ChatGPT in late 2022.
The research team, led by Dmitry Kobak from the University of Tübingen, analyzed more than 15 million biomedical abstracts published between 2010 and 2024. Their goal was to detect patterns in word usage that might indicate the involvement of artificial intelligence in writing these abstracts.
Identifying AI-Generated Text Through Word Patterns
The key discovery is that AI-generated abstracts tend to overuse a specific set of words, including terms like “delves,” “crucial,” “potential,” “significant,” and “important.” These words appeared with far greater frequency after ChatGPT became publicly available.
The researchers compiled a list of 454 words that seem to be favored by chatbots. By tracking the frequency of these words, they estimate that at least 13.5% of biomedical abstracts published recently show signs of having been written with AI assistance.
Variations by Region and Journal Selectivity
Interestingly, the prevalence of AI-generated abstracts is not uniform. In some countries and within certain less selective journals, up to 40% of abstracts may be partly or fully generated by chatbots. This raises questions about the standards and practices across different scientific communities.
Implications for Researchers and Writers
For those involved in science and research writing, these findings highlight the growing presence of AI tools in academic publishing. Understanding these patterns can help editors and reviewers better identify AI involvement and consider ethical guidelines for its use.
As AI writing tools become more common, researchers may want to familiarize themselves with best practices and ethical considerations related to AI-assisted writing. For those interested in gaining practical knowledge about AI tools including chatbots, exploring specialized courses can be beneficial. Resources like Complete AI Training's ChatGPT courses offer insights into how these tools work and how to use them responsibly.
Ongoing Debate in the Scientific Community
The study contributes to ongoing discussions about when and how AI should be used in scientific writing. While AI can aid in drafting and editing, transparency about its use is critical. This research provides a data-driven approach to detect AI involvement, which may influence future publication policies.
Ultimately, the growing role of AI in research writing calls for clear standards to maintain integrity and trust in scientific communication.