
NWO re-examines lottery system for research proposals
The rise of ChatGPT and other chatbots is driving a sharp increase in the number of research proposals submitted to funding agencies worldwide. This is breathing new life into the debate over whether research grants should be awarded by lottery.
Many researchers depend on grants from research funding agencies to advance their careers. As a result, some take a scattershot approach, submitting as many project proposals as possible. Thanks to AI, doing so has become considerably easier.
Is it worthwhile?
At the European Research Council (ERC), the number of applications is rising dramatically, and the same trend can be seen elsewhere. This puts pressure on the experts responsible for assessing the quality of applications. They often do this in their spare time.
Some reviewers are also starting to use AI themselves. This raises the question of how worthwhile the entire process still is. Perhaps the research community needs to find other ways of distributing grants. In Dutch newspaper Trouw, Nijmegen-based researcher Serge Horbach suggests awarding some grants through a lottery.
An old debate
This is not a new debate. Long before AI, critics were already arguing that requiring large numbers of researchers to submit project proposals was a waste of time, given the small chance of securing funding.
Research funding agencies understand this criticism and are trying to improve the process. NWO now initially asks applicants to submit a brief application. Only those who make it to the second round have to write a full proposal. But the easier it becomes to submit applications, the more applications come in.
The middle group
Could a lottery offer a solution? Supporters argue that reviewers can certainly identify the strongest and weakest proposals, but that decisions about those in the middle are already largely a matter of chance. A lottery would be fairer, or at least just as fair, and would save time.
The Rathenau Institute called for a lottery system as early as 2008, followed by De Jonge Akademie in 2014. In 2018, the Netherlands Bureau for Economic Policy Analysis (CPB) made the same recommendation. Two years ago, UMC Utrecht used a lottery to award so-called starter and incentive grants.
At NWO, however, supporters of the idea have consistently received little backing. “We’re waiting to see how experiments in other countries turn out,” then-chair Stan Gielen said in 2017. A lottery is a lazy solution, said current chair Marcel Levi in 2022.
Taking another look
Nevertheless, NWO will revisit the idea of using a lottery in the near future, a spokesperson says. This takes place within a metascientific study, which examines, for example, the impact of research as well as the assessment of research quality.
For now, the increase in AI-generated research proposals in the Netherlands appears to be relatively modest. “We expect that this wave is still coming,” the spokesperson says. “But at the moment, AI is not yet a decisive factor.”
The number of applications submitted to NWO is increasing, but there are other explanations for this as well, including the cuts to research funding introduced by the previous government.
Reason for existence
But introducing a lottery may be easier said than done. Research funding agencies want to reward the best proposals and reject the weakest ones regardless, meaning they still have to scrutinize all applications. In that case, using a lottery for the middle group may not save all that much time.
It could, however, eliminate the appearance of meaningful distinctions between proposals—distinctions that are becoming increasingly difficult to make now that researchers can use AI to quickly put together a reasonably good proposal. At the same time, NWO risks undermining its own reason for existence if it calls into question its mission of funding the best research.
The question, however, is what will happen when the flood of AI-generated proposals eventually reaches NWO as well.
This article was translated using AI-assisted tools and reviewed by an editor.
Discussion