Artificial intelligence will drive changes in research evaluation.

Reshaping Research Funding: How AI Is Forcing an Upgrade of Evaluation Systems

Over the past few decades, the core assumption of research funding systems has been that written proposals can reasonably reflect the quality of research ideas and the capabilities of applicants. The emergence of AI has fundamentally shaken this assumption. As more and more applications are generated with the help of AI tools, funders will find it increasingly difficult to distinguish which contributions come from the researchers themselves and which from the technology they use. When this phenomenon becomes widespread, the screening ability of evaluation systems will face a fundamental logical crisis.

Demand Management Is Only a Stopgap

Faced with a surge in applications, funding agencies first turn to demand management measures: limiting resubmissions, setting application caps, and implementing two-stage review processes. These measures can buy time, but they do not address the core question—in the era of AI, what traits do funding agencies actually hope to identify and support?

Many current funding applications still place heavy emphasis on expected outcomes, even though research itself is full of uncertainty. This incentivizes over-specification and over-promising, and AI further amplifies this tendency by generating more persuasive narratives that align more closely with funder preferences.

Shifting the Focus of Evaluation

A more resilient funding model should pay greater attention to dimensions that are less easily optimized by AI: methodological rigor, relevant capabilities, reflective thinking, and credible pathways to impact. These dimensions require contextual evidence, judgment, and dialogue—they delve into how the research will be conducted and how the researcher will act.

This direction is highly consistent with existing initiatives in research evaluation reform. Alliances such as CoARA and DORA have long proposed that evaluation should focus on both process and output, recognize diverse contributions, and abandon proxy indicators that distort behavior. AI’s role is to push the urgency of translating these principles into funding design to a new level.

In practice, narrative CVs allow applicants to showcase contributions in a more contextualized way, enabling reviewers to make more comprehensive judgments. Competency frameworks provide tools for assessing leadership, collaboration, and reflective thinking. At the institutional level, formative evaluations such as the Dutch Strategic Evaluation Protocol are far less affected by AI-related risks than methods that rely on indicators or large-scale peer review.

Evolution of the Evaluation Process

The evaluation process itself can also be reformed around this new framework. Inviting research administrators to join review panels can provide valuable insights into feasibility and financial management. Including research beneficiaries in reviews helps confirm whether claims about engagement and impact are credible. Interview rounds, though resource-intensive, allow reviewers to gain deep insight into applicants’ reflective ability, collaboration skills, and leadership style—qualities that written text cannot convey.

More importantly, this shift should not be limited to funding applications but should also extend to institutional recruitment and promotion systems.

Reducing the Frequency of EvaluationAnother key issue is the frequency at which individuals are evaluated. A system based on frequent competitive project funding amplifies the distortions brought by AI. Reducing the frequency of individual evaluations—by providing longer-term funding, expanding scholar-fund-based support, or increasing institutional block grants—can help restore balance.

More radical ideas, such as a basic income for researchers, remain controversial, but their value lies in forcing reflection: how much screening is truly necessary, and how much is merely a byproduct of system design rather than an inherent research need?

The Unavoidable Question

A sensitive topic that often arises in internal discussions at funding agencies is whether AI will play a role in the screening or ranking of applications. Avoiding this question is futile, as it is foreseeable that commercial entities are already exploring AI-assisted review solutions. If AI enters this domain, the key issue will be who sets the agenda. The outcome will differ dramatically depending on whether the exploration is led by the academic community on the basis of transparency, accountability, and responsible evaluation principles, or whether it is imposed by profit-driven schemes because funding agencies have no other option.

Conclusion

The disruption brought by AI is often described as a crisis to be managed. But it can also be seen as a forcing function—exposing long-standing flaws and creating political and institutional space for change. Reform initiatives such as DORA and CoARA have already established evaluation directions that prioritize quality, integrity, and inclusivity. The challenge now is to design funding mechanisms that can put these principles into practice in the age of AI.

*This article is based on an opinion piece by Karen Stroobants published in Research Professional News.*