90% of biomedical papers now show signs of AI help

A new analysis reveals that the vast majority of recent biomedical research papers contain language patterns typical of AI writing tools. The finding raises fresh questions about the role of artificial intelligence in scientific publishing. Researchers examined over 1.5 million papers published in 2024. They looked for specific word choices and phrasing that are common in AI-generated text. The results show that about 90% of these papers contain at least some language likely produced by an AI assistant. This does not mean that entire papers are fake. In most cases, researchers likely used AI to edit, summarize, or polish their writing. However, the study’s authors warn that unchecked AI use can blur the line between human analysis and machine-generated content. The trend is growing fast. In 2023, the same team found AI-like language in about 70% of papers. The jump to 90% in just one year suggests that AI tools have become standard in the research workflow. Scientific journals are now facing a dilemma. Many have no clear policy on AI use. Some require authors to disclose it. Others do not. The new data suggests that disclosure is rare, even when AI is used. The study does not claim that AI is making papers less accurate. But it does highlight a new reality: in modern biomedical science, humans and machines often write together. The question is no longer whether AI is used, but how transparent researchers are about it.