Scientists are turning to AI tools more than ever to help them write grant proposals. These proposals are crucial for securing funding for their research careers. However, early data suggests that using AI might steer research toward safer, less innovative ideas.
A recent study found that projects aided by AI showed less originality compared to those written by researchers without AI support. Interestingly, these AI-assisted proposals had a slightly higher chance of being funded. Misha Teplitskiy, a researcher from the University of Michigan, noted this trend could lead to a lack of diversity in scientific innovation. “We could be on a path towards homogeneity,” he remarked, highlighting concerns about the future of research creativity.
The rise of large language models, like ChatGPT, has changed how scientists approach their work. A 2024 survey showed that many researchers began using AI tools in 2023. Dashun Wang and Yifan Qian, researchers from Northwestern University, sought to understand the effect of this rapid adoption on the quality of funded research.
They analyzed thousands of grant proposals submitted to the National Science Foundation (NSF) and the National Institutes of Health (NIH) from 2021 to 2025. By comparing older, human-written abstracts with those generated by AI, Wang and Qian developed a method to identify the usage of AI in grant writing.
Their findings showed a significant spike in the use of AI tools for grants starting in early 2023, coinciding with the public release of ChatGPT. However, not every scientist embraced this technology; some deliberately avoided using AI. Those who did use it typically generated about 10-15% of their proposal text with AI assistance.
The effects varied by funding agency. At the NIH, high AI involvement in proposals was linked to a 4% increase in the chances of winning funding. The funded projects also produced 5% more published papers, although these were often less cited than others. Conversely, the NSF showed no clear benefit from AI usage.
This connection between AI use and funding rates, particularly at the NIH, raises important questions about the future of scientific research. As AI becomes more integrated into the funding process, it’s essential to balance innovation with the efficiency that these tools provide.
For more insights and statistics on AI in research, check out this report from the National Institutes of Health here.
Source link
Funding,Machine learning,Science,Humanities and Social Sciences,multidisciplinary
