In het Benelux secretariaat te Brussel zal op vrijdag 22 mei 2026, WB-NSC onderzoeker Jeroen Zandberg een Engelstalige presentatie houden over de lopende onderzoeken die het Wetenschappelijk Bureau NSC heeft naar AI, goed bestuur en regionale identiteit. Deze presentatie is onderdeel van het RISE programma van het Leuven Centre for Global Governance Studies.
Hieronder staat een voorbeschouwing van de presentatie:
From Language Rights to Language Models: Benelux Integration through AI-Driven Multilingual Governance
The Benelux has great potential as a multilingual AI ecosystem. Beyond Dutch, French and German, the region includes several recognized regional and national minority languages that are currently underrepresented in digital infrastructures. My research shows that AI already makes it technically feasible to integrate these languages into administrative processes at relatively low cost.
In my work at the Scientific Bureau of Nieuw Sociaal Contract (NSC), I investigate how multilingualism and artificial intelligence are organized within Dutch local and regional public administration. In multiple recent research projects on the use of officially recognized national minority languages which have limited support in government administration, Frisian, Limburgish and Stellingwerfs (Low Saxon), I observed that the way multilingualism is implemented depends on an interplay between various factors, like the specific legal status, technological development of the language and the public administration and the priorities of policy makers. I am currently also conducting research into how selected Dutch municipalities organize and govern the use of AI, revealing both experimentation and fragmentation at the local level.
I argue that the combination of multilingual governance and AI constitutes an opportunity for Benelux cooperation. Drawing on empirical findings from these studies, I will show how current challenges—such as limited datasets for low-resource languages, fragmented administrative approaches, and the absence of shared standards—can be addressed more effectively on a supranational, Benelux, scale.
Three concrete opportunities for Benelux cooperation emerge. First, pooling public-sector language data can significantly improve AI performance for both national and regional languages, overcoming limitations that individual countries face. Secondly, the development of shared standards for terminology, translation quality and human-in-the-loop processes can reduce duplication and increase reliability in multilingual service delivery. Third, coordinated investment in public-value-oriented language models, building on initiatives such as GPT-NL (the Dutch LLM for public administrations), and strengthen digital sovereignty while achieving economies of scale.
Finally, I will connect insights from language policy, public administration and AI governance, in order to highlight how institutional choices, such as procurement, data management and organizational capacity, shape the success of multilingual AI. By bringing these perspectives together, this presentation aims to contribute to a broader interdisciplinary agenda in which the Benelux can become a vehicle for inclusive and scalable digital governance.