The Dutch AI strategy is overlooking its major cultural opportunity

"”The question is no longer simply whether libraries use these tools, but whether they have a say in how the tools work.”
Wilma van Wezenbeek, KB National Library

When the Dutch government announced its International AI Strategy When the government presented its plan, the ambition was clear. According to the government, artificial intelligence is no longer just an ordinary technology, but a strategic factor that will shape our economy, security, democracy and international standing. The Netherlands must therefore invest in computing power, high-quality data and European technological autonomy. According to the government, AI is a key technology for our future values, prosperity and resilience.

"AI is a key technology for our future values, prosperity and resilience."

It is a compelling narrative. But anyone who reads the document carefully will notice a striking omission. Whilst the strategy constantly refers to data, public values and European autonomy, the cultural sector remains almost entirely overlooked. This is remarkable, because it is precisely in Europe that a fundamentally different debate has begun. There, culture is no longer seen as a sector that must adapt to AI, but increasingly as an indispensable building block of the AI infrastructure itself.

From cultural policy to AI infrastructure

This shift is perhaps most clearly evident in the recently published European Books Data Commons- a study by the Royal Library and Europeana.

The report describes an ambitious European initiative to make millions of digitised books from national libraries available for the development of European AI models. Not because libraries suddenly want to become technology companies, but because they realise that the question is no longer simply whether libraries use AI. The fundamental question is who determines how AI represents knowledge, which sources are made visible, and what values are embedded within them. 

That is a fundamentally different way of thinking about culture. Libraries no longer merely preserve the past; they provide the raw materials for the knowledge infrastructure of the future. The study also makes it clear that Europe possesses something that commercial entities can scarcely replicate: millions of books in dozens of European languages, carefully documented, historically rich and of high quality. It is precisely this multilingualism and cultural diversity that are in short supply in today’s AI market, which is dominated by English-language commercial datasets. 

"As AI becomes the dominant interface to knowledge and culture, The next step is clear: make that data AI-ready." — Harry Verwayen, Europeana.

Interestingly, the Council for Culture had also previously taken the same approach. In its advice Artistic rather than artificial The Council does not describe AI primarily as a technological issue, but as a cultural development.

"AI is not just a technological development, but is changing the way culture is created, shared and experienced."

After all, AI learns from books, paintings, music, films, photographs and other cultural expressions produced by generations of creators. At the same time, AI is also changing the way in which new culture is created, disseminated and appreciated. The Council therefore advocates that the cultural and creative sector should not be treated as an afterthought in AI policy, but as a full partner in the development of public AI infrastructure. 

It is precisely against this backdrop that the Dutch AI strategy appears surprisingly traditional. Understandably, it focuses on chips, cloud infrastructure, language models, investment and international competitiveness, but pays scant attention to the institutions that perhaps manage Europe’s most important public data sources. This is more than a symbolic omission. It means that a large part of the Netherlands’ cultural capital remains outside the scope of the strategic analysis.

The data that teaches AI to think

This has long since ceased to apply solely to libraries. Dutch museums collectively hold millions of objects that have been digitised over the past few decades. Archives manage centuries’ worth of historical documents, maps and audiovisual material. Image and sound collections contain unique recordings of the Dutch language, culture and history. It is precisely these collections that constitute high-quality, curated and reliable datasets of great value to European AI systems. Unlike ephemeral internet data, they are carefully described, tagged with metadata and embedded within professional knowledge structures. This makes them not only culturally valuable, but also technologically interesting.

The discussion is thus shifting, almost imperceptibly, from cultural policy to strategic data policy. Whereas in recent years the focus has mainly been on copyright and the question of whether AI poses a threat to artists, a much more fundamental question is now emerging. If AI increasingly determines what knowledge people are exposed to, what stories are told and which languages remain visible, who then manages the sources from which that AI learns? The answer is increasingly: libraries, archives, museums and other heritage institutions.

A missed strategic opportunity

That is precisely why it is striking that the Dutch AI strategy barely reflects this development. Whilst Europe is increasingly viewing culture as a strategic resource for AI, the Dutch approach remains largely stuck in a traditional dichotomy between technology policy and cultural policy. This is a pity, because the Netherlands actually possesses an exceptionally strong cultural infrastructure. Our museums are among the best in the world, and the Royal Library plays a pioneering role internationally, Europeana is managed from the Netherlands, and our heritage institutions hold collections of global significance. It is precisely these institutions that Europe is now seeking to mobilise for the development of public AI.

Perhaps that is the most important lesson to be drawn from the three recent documents. The international AI strategy sets out how the Netherlands intends to remain a technological leader. The Council for Culture makes it clear that AI is inextricably linked to culture. The European Books Data Commons- The study then takes the next step and shows how cultural institutions themselves can become strategic AI infrastructure. It is precisely these three insights that deserve to be read in conjunction with one another. For if Europe truly wishes to chart its own course in AI, it will have to invest not only in the computers on which AI runs, but also in the books, images, stories and collections from which AI learns to think.

The Netherlands is investing billions in AI, but may well be overlooking its greatest strategic asset. Whilst Europe is increasingly viewing libraries, museums and cultural heritage as the cornerstone of public AI, the Dutch cultural sector remains conspicuously absent from the national AI strategy. Why is this a missed opportunity?

Appreciate this article!