The Netherlands Has Been Training AI Experts Since 1988 — Long Before the ChatGPT Era
Amsterdam, Thursday, 28 May 2026.
When Utrecht University launched its first AI degree program in September 1988 with just 69 students, few could have predicted today’s generative AI revolution. That 38-year head start raises a compelling question: did those pioneers gain a lasting advantage?
A Program Born From Philosophy, Not Computer Science
The story of Dutch AI education does not begin in a technology lab or a computer science department. In September 1987, academics Jan Bergstra and Gerard Renardel de Lavalette launched the initiative for the program from within the faculty of philosophy at what was then known as the Rijksuniversiteit Utrecht — today’s Universiteit Utrecht [1]. The choice of faculty was deliberate and telling. Computer science programs at the time were embroiled in a struggle for academic recognition, while the philosophy faculty had the institutional flexibility to experiment [1]. As Bergstra himself recalled of the interdisciplinary tensions of that era: ‘Bij wiskunde zeiden ze bijvoorbeeld: een informaticus is een wiskundige die overdag ook tv kijkt’ — loosely translated, mathematicians dismissed computer scientists as merely mathematicians who also watched television during the day [1].
Pioneers in an Underdeveloped Field — By Definition
The realities of studying AI in the late 1980s were far removed from the polished, resource-rich programs that exist today. With no dedicated curriculum in place, students often joined existing courses from other disciplines rather than attending classes built specifically for them [1]. Zijlmans noted plainly: ‘Er kwamen geen nieuwe vakken voor onze studie, we deden mee met andere studies’ [1]. Attrition was steep: of the 69 students who began in September 1988, only 40 remained by December 1 of that same year, and by 1989 only 10 had passed their propedeuse — the Dutch first-year qualifying examination [1]. That means 14.493 percent of the original intake cleared the first formal academic hurdle.
Graduation, the AI Winter, and a Long Wait
In 1992, Johan Caljé and René Westenberg — both now 61 years old — became the first people in the Netherlands to graduate with a degree in artificial intelligence [1]. The timing, however, proved deeply unfortunate. Shortly after that milestone, the so-called ‘AI winter’ set in during the early 1990s, a period in which enthusiasm for artificial intelligence collapsed and employment opportunities in the sector almost entirely disappeared [1]. The gap between that pioneering graduation year of 1992 and the mainstream resurgence of AI interest is stark: it would take another three decades before the field truly entered public consciousness, marked most visibly by the launch of ChatGPT on November 30, 2022 [1]. That is a gap of 30 years between the Netherlands’ first AI graduates and the moment the technology became a household conversation.
From Utrecht’s Lecture Halls to a National AI Education Ecosystem
The publication of the de Volkskrant retrospective on May 28, 2026 coincides with a moment of genuine reflection across the Dutch education sector [1]. In late April 2026, the original CKI students and staff reunited in Utrecht to discuss the rise of AI and the impact of ChatGPT — a reunion that brought together the field’s founding generation with the perspective of hindsight [1]. Bergstra, now an emeritus professor of applied logic, offered a measured assessment of the current AI landscape: ‘Vooral dat het vanuit de neurale netwerken komt, dat had ik in de jaren tachtig niet voorzien. Het is de beschikbaarheid van brute rekenkracht, die we vroeger niet hadden, die deze technologie mogelijk maakt’ [1]. His observation that it is raw computational power — unavailable in the 1980s — that has made modern AI possible is a point now broadly accepted across the field [GPT]. He added a note of caution regarding AI replacing humans: ‘Wat daarvoor nodig is, is nog volkomen onbekend terrein’ [1].
Building Bridges Between MBO and HBO in the North
In the north of the Netherlands, a parallel effort is taking shape at an institutional level. On May 27, 2026 — just one day before the Volkskrant article was published — Mark Tiele Westra was appointed to a dual role as Practor Toegepaste AI (Applied AI) at Firda and as Associate Lector Applied AI at NHL Stenden [6]. The appointment is specifically designed to bridge the gap between MBO (vocational) and HBO (higher professional) education in the region [6]. Westra brings an unusually broad background to the role: he holds a doctorate in Theoretical Physics from TU Eindhoven, a master’s degree in Philosophy of Science, Technology and Society from TU Twente, and more than 15 years of experience in software development, machine learning, large language models, and AI systems [6].
A European Policy Framework Shapes the Classroom
The Dutch educational push on AI does not exist in isolation. At the European level, the institutions of the EU have actively sought to shape how AI is introduced into classrooms. The European Commission has published updated ethical guidelines for educators on the use of artificial intelligence and data in teaching and learning, issued under Action 6 of the Digital Education Action Plan [2]. The guidelines are directed primarily at teachers and educational staff in primary and secondary education, and were updated in response to three converging pressures: the sharp rise in AI use in education; the introduction of binding legislation in the form of the EU AI Act, which must be carefully observed; and a growing need for ethical and critical AI literacy [2].
What the Pioneers Tell Us About the Present
The May 28, 2026 retrospective in de Volkskrant ultimately poses a question that cuts to the heart of AI education today: did those first graduates actually benefit from their head start? The evidence is ambiguous. Some, like Bas Haring, built careers that placed them at the frontier of AI’s latest wave [1]. Others, like Ronald Lemmen, found that the AI winter rendered their expertise commercially redundant for years, diverting them into entirely different fields [1]. Esther Zijlmans offered perhaps the most grounded assessment: ‘Als je het mij vraagt, is de komst van generatieve AI, zoals ChatGPT, een net zo grote doorbraak als de komst van het internet’ [1]. That comparison — generative AI as the new internet — is one that carries significant weight coming from someone who was studying artificial intelligence when the World Wide Web itself did not yet exist [GPT].