Research funders are facing a new problem as artificial intelligence spreads into the grant review process. A senior manager told a European conference that AI is not yet making the job of assessing funding bids any easier.
Roisin Cheshire, head of individual-led research at Research Ireland, spoke at the 2026 European University Association Funding Forum in Brno, Czech Republic, on October 15. She said there is not enough evidence to show that AI makes grant reviews easier overall.
Her comments come as funders handle a sharp rise in applications. Data from the UK-based Research on Research Institute shows grant applications across 12 major European funders were 57% higher on average in 2025 than in 2022.
ChatGPT launched in 2022.
Funders Report AI-Generated Reviews
Cheshire said managing reviews written by AI has become an extra challenge for her agency.
“As I think everybody is experiencing, the numbers of applications are just going up and up,” she said. She added that Research Ireland has rules on AI use but is “struggling with that a little bit.”
“We’re starting to see some reviews come in which are very obviously AI-generated,” Cheshire said. “How do you police that, and how do you say to that reviewer, we can’t use the review?”
She said many agencies are testing pilot schemes. However, she said the “proof” that AI can simplify assessments is “not there yet.”
Research Ireland has set up an AI governance community. It is also writing a dedicated policy and watching how other funding agencies handle the issue.
“So we are on it,” Cheshire said, “but it’s certainly not making my life any easier just yet.”
Panellists Stress Human Oversight
Other speakers said people must stay involved in evaluating grants and projects. Jean-Pierre Vidal, director of the EIB Institute at the European Investment Bank, said “AI is…something we cannot ignore.”
Vidal said many of the bank’s reviews require staff to meet people “on the ground” to discuss proposals in detail. He said AI cannot do this part of the work.
Bank staff are encouraged to use AI tools. Vidal said these tools work best for fixing typos, removing repetition and improving the structure of writing, while humans still handle the main evaluation.
Jonna Korhonen, director for higher education policy at Finland’s Ministry of Education and Culture, said AI has pushed questions about quality and assessment methods to the front across higher education. She called for evaluation frameworks to be rethought so assessment becomes “more comprehensive.”
In a separate session, speakers discussed how to keep trust in university assessment as AI use grows. Alex Usher, president of Canada’s Higher Education Strategy Associates, said AI “makes assessment really difficult in universities [and] very expensive.”
Usher said universities “are going to have some choices to make” and must ensure degrees stay valuable to students and employers. “Can universities guarantee students that their efforts are actually going to be assessed fairly?” he asked.
