One programme officer in the United States recently told the head of the Technology Association of Grantmakers that they had received 30 nearly identical grant proposals.
It’s a small story, but it points to something funders are starting to feel: the written application is no longer working. The Chronicle of Philanthropy covered this this week, and I have been reading it alongside a few other pieces that look at the same problem from different angles. Together they suggest this is bigger than a busy funding round, and that funders in Aotearoa have some choices to make about what comes next.
What we know so far
The strongest numbers come from research funding. An analysis by two UCL professors looked at 12 research funders across six countries and found that applications rose 57% between 2022 and 2025, with one European fellowship scheme up 142%. In philanthropy the picture is more anecdotal. The Chronicle describes a room of 60 grant makers being asked whether they were seeing more applications, and every hand went up… although nobody had hard data to back it.
The same thing is happening outside funding. The Guardian reported this week that UK councils are struggling with AI-written complaints, which used to be two pages and now closer to 20. Every one of them still needs a person to respond.
None of this data is from Aotearoa but the pattern is consistent enough that I would not wait for perfect data.
Writing got cheap, and reading did not
The best explanation I have read comes from UK fundraiser Alex Evans, who argues that application processes have always relied on an imbalance. Writing a good application took real time and skill, and that effort acted as a filter which kept the volume manageable for the people reading. Now anyone can produce a polished application in minutes, but reading one carefully takes as long as it ever did.
The old filter was never fair. It favoured organisations with paid fundraisers and confident written English, and many of us have spent years asking funders to make applying easier. AI has done exactly that, and maybe that difficulty was holding the system together.
Three ways funders could respond
Use AI to read. The UCL authors argue that banning AI is unenforceable, and that funders should use it to assess applicants' track records instead of their prose. Some foundations already use AI as a thinking partner, with people reading every proposal at the start and making the decision at the end. The risk is a loop where machines write for machines, and where judging on track record rewards those who have already been funded
Ask for less. One nonprofit technology leader told the Chronicle that if staff only have time to read a paragraph, a paragraph is all they should ask for. The Technology Association of Grantmakers' framework makes a similar point, which is that if the process is too heavy, changing the process is cheaper than adding AI to both sides of it.
Lean on relationships. Shorter forms and more conversations sound like what the sector has been asking for. Evans' warning is that this can slide back into funding who you know, and the organisations at the front of that queue are rarely the small, new or marginalised ones.
The question underneath
At Fundsorter we build tools that help community organisations find the right funders, so we are part of this story. I think about our role in it a lot. My view is that the answer is transparency. It comes before the application. If funders are clear and public about what they fund, and organisations can see quickly whether they are a good fit, fewer people spend time writing applications that were never going to succeed and fewer funders spend time reading them.
That still leaves a harder question for funders: if a well-written application no longer tells you much about the organisation behind it, what do you want to know instead, and how will you find it out in a way that is fair to the groups you have not met yet?
