What do we really know about the technology behind ChatGPT and other generative AI? Researchers Mark Dingemanse and Andreas Liesenfeld of the Centre for Language Studies (CLS) keep close track of developments in AI. Their project, which provides insight into the transparency of AI systems, has been awarded an Impact Explorer 2025 grant by NWO and has been nominated for the Huibregtsen Prize 2025.
They use the grant to explore the societal impact of the European Open Source AI Index: a dynamic overview of the transparent, accessible, and reusable AI projects in Europe. The index, which was launched in 2023, helps researchers, policy-makers and educational institutions to find their way in an ever-growing landscape of open technology that provides an alternative to the closed systems of Big Tech. “If we want public investment to serve public interest, then it is crucial to invest in open and transparent AI ecosystems,” according to Dingemanse.
European Open Source AI Index
Generative AI seems to have become commonplace in a short period of time, but its most commonly used variants are by no means transparent. In the European Open Source AI Index, Dingemans and Liesenfield map out the rapidly changing landscape of generative AI in an accessible manner. They make crucial information about training data and algorithms available for both science and society.
The index, at osai-index.eu, presents a ranking of AI systems, extensive search options and practical guidelines. Their work is cited in scientific publications and policy papers by, among others, the UN, the UK House of Lords, Deutsche Bank and the Dutch government. The index provides a nuanced alternative to the simplistic binary ‘open versus closed’ by measuring openness based on 14 indicators. Their analysis shows that transparency and innovation are not mutually exclusive but can instead strengthen one another.
From research to education and policy
The researchers will be using the Impact Explorer grant to develop open source tools that contribute to critical AI literacy in education and policy-making. In doing so, they collaborate with the National Education Lab AI (NOLAI) and with iHub, Radboud University’s interdisciplinary research institute.
School pupils and students learn, for instance, how AI models work and how they can question them critically. Policy-makers are given actionable information that they can use to audit and question the use of AI in public domains such as healthcare, education, and government.
Nominated for the Huibregtsen Prize 2025
The project has been nominated for the Huibregtsen Prize 2025, a prestigious award for research that combines scientific quality and innovation with societal value. The jury selected six projects out of 28 candidates. Each year, a minimum of four and a maximum of six projects are nominated; only one of these will ultimately receive the prize.
The jury was impressed not only by the speed at which this project has generated significant scientific impact, but also with the way in which the team of relatively young scientists is actively committed to the societal impact of their work.
The Huibregtsen Prize was established by Foundation Evening of Science & Society in 2005 and is named after founder Mickey Huibregtsen. The award ceremony will take place on the yearly Evening of Science and Society on Thursday 16 October.