In an era where artificial intelligence is rapidly transforming various sectors, including research and academia, two researchers from University College London have raised concerns about the challenges and potential pitfalls AI poses to the existing research funding system. Professors Geraint Rees and James Wilsdon argue in a recently published analysis in Nature that the current frameworks for allocating billions in research grants need an urgent overhaul to keep pace with the capabilities of “agentic AI.”
Understanding Agentic AI and Its Impact
Agentic AI refers to the sophisticated capabilities of AI tools, particularly those that extend beyond single-task responses. These AI agents can autonomously navigate tasks across multiple steps, such as searching the internet, analyzing documents, and even drafting complex outputs like coding or comprehensive grant proposals. When applied to academic funding, these AI tools can rapidly generate high-quality grant applications by synthesizing a researcher’s previous work and adapting to proven successful proposal formats.
This revolutionary ability presents a dual-edged sword: while it democratizes access to sophisticated proposal drafting, it could overwhelm funding agencies with an unmanageable influx of applications, making the selection process more arbitrary and less about distinguishing nuanced, groundbreaking ideas.
Evidence of a System Under Strain
The UCL researchers highlight data supporting their concerns. A survey revealed that from 2022 to 2025, there was a dramatic surge in grant applications, with increases as high as 142% for specific fellowships. This spike coincides with the introduction and advancement of large language models like ChatGPT, suggesting a direct correlation between AI adoption and increased application volumes.
This trend poses significant challenges for peer reviewers and funding bodies, who may soon face avalanches of proposals generated with or reviewed by these very AI systems. The current manual processes in place may buckle under such pressure, leading to systemic inefficiencies.
Recommendations for a New Strategy
Rather than attempting to restrict AI usage—which could be both impractical and counterproductive—Rees and Wilsdon recommend leveraging agentic AI to redesign the funding process itself. They advocate for AI tools to assess and prioritize applications more holistically, focusing on factors like consistency, novelty, and potential impact. This technological pivot would help mitigate selection biases, ensure transparency, and support inclusivity, particularly for emerging researchers or institutions with less prestige.
Conclusion: Navigating the Future Wisely
As we advance into 2026 and beyond, the necessity to adapt our funding systems in light of AI’s capabilities has become imperative. The insights from Rees and Wilsdon encourage stakeholders to not only acknowledge the strains current systems face but to proactively harness AI’s potential to enrich and sustain a fair research funding ecosystem. Without strategic adaptation, we risk diluting excellence amid overwhelming choices, challenging the very progress and innovation the funding aims to foster.
By rethinking and updating our approaches to research funding, we can better align with the evolving technological landscape, fostering an environment where innovation and excellence can truly thrive.