The Rise of Large Language Models and the Direction and Impact of US Federal Research Funding

TL;DR

This study examines how the rise of large language models reshapes the US federal research funding landscape by linking LLM usage in grant abstracts to idea positioning, funding success, and downstream outputs. It combines confidential proposal data with the full population of NSF/NIH awards and uses a distributional LLM-detection framework to estimate the LLM involvement parameter $\alpha$ and SPECTER2-based semantic distinctiveness. Findings show a rapid, bimodal uptake of LLMs starting in 2023, with higher $\alpha$ consistently pulling ideas toward the recent funding frontier; agency-dependent effects reveal NIH proposals with higher $\alpha$ have higher funding probability and more publications (primarily non-hit papers), while NSF shows no such advantages. The results imply that generative AI is altering portfolio composition and governance, potentially enhancing writing efficiency while risking reduced exploration, underscoring the need for policy and governance mechanisms to preserve portfolio diversity and accountability.

Abstract

Federal research funding shapes the direction, diversity, and impact of the US scientific enterprise. Large language models (LLMs) are rapidly diffusing into scientific practice, holding substantial promise while raising widespread concerns. Despite growing attention to AI use in scientific writing and evaluation, little is known about how the rise of LLMs is reshaping the public funding landscape. Here, we examine LLM involvement at key stages of the federal funding pipeline by combining two complementary data sources: confidential National Science Foundation (NSF) and National Institutes of Health (NIH) proposal submissions from two large US R1 universities, including funded, unfunded, and pending proposals, and the full population of publicly released NSF and NIH awards. We find that LLM use rises sharply beginning in 2023 and exhibits a bimodal distribution, indicating a clear split between minimal and substantive use. Across both private submissions and public awards, higher LLM involvement is consistently associated with lower semantic distinctiveness, positioning projects closer to recently funded work within the same agency. The consequences of this shift are agency-dependent. LLM use is positively associated with proposal success and higher subsequent publication output at NIH, whereas no comparable associations are observed at NSF. Notably, the productivity gains at NIH are concentrated in non-hit papers rather than the most highly cited work. Together, these findings provide large-scale evidence that the rise of LLMs is reshaping how scientific ideas are positioned, selected, and translated into publicly funded research, with implications for portfolio governance, research diversity, and the long-run impact of science.

Figures (23)

• Figure 1: Rapid rise and bimodal distribution of LLM use in US federal research funding. (a-d) Corpus-level estimates of LLM use ($\alpha$) for private and public NSF and NIH grants from 2021 to 2025, computed using rolling three-month windows (points). Solid lines show locally weighted regressions. The vertical dashed line marks November 30, 2022, corresponding to the public release of ChatGPT. (e-h) Distributions of individual-grant $\alpha$ for private and public NSF and NIH proposals and awards with start dates between 2023 and 2025, showing a bimodal pattern consistent with a split between minimal and substantive LLM use across grants.
• Figure 2: LLM use and semantic distinctiveness in US federal research funding. (a-d) Regression estimates relating grant-level LLM use ($\alpha$) to semantic distance from abstracts funded in the prior year within the same agency, expressed as within-year percentiles. Panels show results separately for private NSF (a), private NIH (b), public NSF (c), and public NIH (d) grants. All regressions include grant start year, field, and investigator fixed effects, as well as controls for funding amount. Points indicate coefficient estimates, and bars denote 95% confidence intervals. Negative coefficients correspond to proposals and awards that are positioned closer, in semantic space, to recently funded work within the same agency.
• Figure 3: LLM use and federal research proposal success. Based on private NSF and NIH proposal submissions from two large US R1 universities, this figure examines the relationship between LLM use at submission ($\alpha$) and proposal success. (a) Regression estimates for NSF submissions. (b) Corresponding estimates for NIH submissions. All regressions include proposal request start year, field, and investigator fixed effects, as well as controls for requested funding amount. Points indicate coefficient estimates, and bars denote 95% confidence intervals.
• Figure 4: LLM use and federal research funding outputs. (a-b) Regression estimates relating grant-level LLM use ($\alpha$) to the total number of resulting publications for NSF (a) and NIH (b) grants. (c-d) Corresponding estimates for high-impact outputs, where a "hit" paper is defined as one whose citations fall within the top 5% of all papers published worldwide in the same year and field. All regressions include grant start year, field, and investigator fixed effects, as well as controls for funding amount. Points indicate coefficient estimates, and bars denote 95% confidence intervals.
• Figure S1: LLM use and downstream output in US federal research funding. (a-b) Regression estimates relating grant-level LLM use ($\alpha$) to high-impact outputs, where a "hit" paper is defined as one whose citations fall within the top 1% of all papers published worldwide in the same year and field. All regressions include grant start year, field, and investigator fixed effects, as well as controls for funding amount. Points indicate coefficient estimates, and bars denote 95% confidence intervals.