GoKawiilStudy: GPT-4o grant screening favored conventional proposals over novel ones
Diogo Machado of consultancy Technopolis tested OpenAI's GPT-4o on 139 real grant proposals submitted to Austria's Science Fund in 2021, asking it to pick the better of each pair. The model matched human reviewers' funding decisions 60% of the time, but was notably less likely than humans to back proposals rated high for novelty, according to the study published in Scientometrics.
GoKawiil's interpretation of the reporting above, not reported fact.
The findings suggest that relying on AI tools to screen grant applications could systematically disadvantage unconventional research ideas, even as funders explore such tools for speed and efficiency. If adopted at scale, this bias could skew science funding toward incremental work and away from potential breakthroughs, though the study examined only one model and one funder's proposals.
- GPT-4o agreed with human grant reviewers in 60% of 139 paired comparisons.
- The AI was systematically less willing to fund proposals rated high in novelty.
- Proposals came from Austria's Science Fund 2021 grant cycle, including human scores on novelty, quality, team capacity and feasibility.
Source: nature.com — Yates, 2026-11-03
Published there as: “Will AI grant reviews make science more conservative?”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.