Could agentic AI topple grant-funding systems?

2026-04-27 · Source: Machine learning : nature.com subject feeds · Field: Science & Research — Artificial Intelligence & Machine Learning, Research Methodology & Innovation · Depth: Intermediate, medium

Summary

The increasing use of AI agents in grant writing is transforming the research funding landscape, leading to a sharp rise in application volume and perceived quality, which threatens the current evaluation system. AI agents, powered by large language models, can generate, review, and submit grant applications with minimal human intervention by planning and executing sequences of steps. Data from 12 multidisciplinary funders across seven countries and the EU, including the Australian Research Council and Wellcome, show application increases ranging from 14% to 142% between 2022 and 2025. A 2025 Elsevier survey found 58% of researchers use AI tools, with 41% drafting grant proposals. This surge in AI-optimized proposals makes it harder for funders to discriminate between submissions, potentially leading to arbitrary funding decisions and a system that evaluates AI simulation rather than novel ideas.

Key takeaway

For AI Scientists involved in grant applications, the rise of AI agents necessitates a strategic shift. Relying solely on AI-generated proposals risks contributing to a dysfunctional system where true innovation is obscured. You should focus on developing unique research narratives and demonstrating a strong, verifiable track record, as funders are moving towards evaluating investigators and teams rather than just proposal text. Consider how your work can genuinely stand out beyond AI optimization.

Key insights

AI agents are increasing grant application volume and quality, challenging traditional funding evaluation systems.

Principles

• AI agents optimize for outcomes, not just craft.
• Bans on AI use in grants are difficult to enforce.
• High-quality AI-generated proposals can overwhelm reviewers.

Method

Funders should shift evaluation emphasis from written proposals to principal investigators, research teams, and their track records, using interviews and portfolio-based assessments.

In practice

• Implement track-record verification for applicants.
• Conduct interviews for shortlisted grant applicants.
• Assess sustained team performance via portfolios.

Topics

• Agentic AI
• Grant Funding Systems
• Research Grant Applications
• Large Language Models
• Research Evaluation

Best for: AI Scientist, Research Scientist, Policy Maker, Consultant

Related on AIssential

• Investing in multi-agent AI safety research — Multi-agent AI systems require new safety research to manage emergent collective behaviors and systemic risks.
• The AI scientist: now academic papers can be fully automated, what does this mean for the future of research? — Agentic AI systems now autonomously plan, execute, and iterate, transforming research and creative industries.
• It’s the Lessons We Learned Along the Way. Or, Is It? — AI agents can efficiently complete complex data tasks, but may diminish human understanding and critical thinking development.
• In Defense of Tokenmaxxing — Experimentation with AI token consumption is crucial for enterprise adaptation to agentic AI, despite potential for misuse.
• Is it time to ‘cap and trade’ credits for research-funding proposals? — AI-generated grant proposals risk overwhelming research funding systems, necessitating systemic solutions.
• Responses to the AI grant flood must prioritize fairness as part of excellence — AI-driven application surges demand grant system reform to ensure fairness and maintain funder credibility.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.