Researchers face record-low chances of getting a grant partly due to rise of AI

Research grant providers are receiving far more applications while the number of awards remains largely fixed, pushing researchers’ odds of winning funding to record lows. The European Research Council’s 2027 Starting Grant attracted 7,710 early-career scientists, but only 373 will receive funding - a success rate under five percent. The Marie Curie program received 21,627 applications this year, almost 27 percent more than last year. The Dutch research funder NWO says award rates are declining and expects at least one program’s rate to fall below 10 percent, despite aiming for 20 to 25 percent because applications consume substantial researchers’ time.

Researchers studying the funding system say generative AI is contributing to the surge by making it easier to produce polished proposals quickly. Serge Horbach, who assesses applications for a Flemish grant provider, reports that submissions now tend to be well written and visually coherent, making it harder to screen out weak proposals on basic presentation. But polish does not necessarily reflect the applicant’s grasp of the research: some candidates struggle in interviews to explain how they would carry out their projects. Their interviews can take opportunities from stronger applicants. The result is a growing mismatch between the volume and appearance of applications and the limited number of credible, fundable projects, raising concerns that the application process is becoming wasteful and unsustainable.