AI-assisted grant proposals may win more often—while narrowing research ideas
A new study from Northwestern's Kellogg School of Management found that research proposals showing stronger signs of AI-assisted writing were four percentage points more likely to receive funding from the National Institutes of Health (NIH).
A Northwestern University study reveals that research proposals utilizing artificial intelligence (AI) to draft their content have a 4-percentage-point higher likelihood of securing funding from the National Institutes of Health (NIH). However, these AI-assisted proposals also exhibit less semantic distinctiveness compared to previously funded projects, potentially narrowing the scope of scientific research.
The research, led by Dashun Wang and Yifan Qian, highlights a trade-off between increased funding success and a possible decline in research innovation. Wang, a professor at Kellogg School of Management, suggests that AI could steer funding toward safer and more conventional research ideas.
The study, published in the Proceedings of the National Academy of Sciences, examines the impact of large language models (LLMs) on US federal research funding. With the public release of ChatGPT in late 2022, LLM usage in grant writing skyrocketed, leading to a bifurcation in proposal writing styles. Some proposals showed minimal AI assistance, while others relied heavily on AI.
At NIH, proposals with stronger AI involvement had a higher chance of receiving funding and produced more follow-on publications, although not more highly cited papers. This suggests AI might enhance research productivity without necessarily leading to breakthrough discoveries. In contrast, at the National Science Foundation (NSF), no significant relationship was found between AI use and either funding success or publication output.
The disparity between NIH and NSF results prompts a broader discussion about the future of scientific discovery. The study's authors argue that higher LLM involvement is tied to lower semantic distinctiveness, which could reflect a shift towards incremental, executable projects that align with reviewer expectations. However, this trend might also limit exploration of unconventional ideas, impacting public trust in the stewardship of taxpayer-supported science.
Written by urgent.news from Phys.org's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
