Scientists who lean more on AI tools to write grant applications tend to pitch less original ideas that look more like what was funded the year before. Yet at the NIH, these AI-assisted proposals were more likely to win funding, and led to slightly more papers but not more highly cited ones.
A new study of U.S. federal grant applications suggests that scientists using generative AI tend to pitch research closer to previously funded topics. At the NIH, AI-heavy proposals won funding more often and preceded modestly more routine papers.