Student interest in information and communication technology programs has fallen as prospective students question whether traditional coding education is necessary in the era of artificial intelligence.
Data requested by EenVandaag shows a decline of more than 25 percent in new enrollments over the past three years.
In 2022, more than 10,000 students started an ICT program at universities or universities of applied sciences. By last year, that number had dropped to 7,600. Some students now view AI as a replacement for human programming skills.
“My brother first wanted to study computer science—just like me—but then changed his mind,” said Renske, a master’s student in computer science at Leiden University. “He said: ‘You see many people using AI now, so why would I learn to code?’ He saw more job opportunities in another field.”
Felienne Hermans, a professor of computer science education at Vrije Universiteit Amsterdam, understands very well that students struggle with the question of why they would still study this. “Many companies choose to use AI agents instead of programmers,” Hermans said.
Despite this trend, demand for ICT professionals remains high.
A recent report from the staffing agency Randstad indicates that 86 percent of companies are actively searching for technical talent. Hermans characterizes this as the “paradox of AI.”
“It is sold as something that saves, but if AI generates a lot of text, photos, or code, you need more people to check it all,” she said.
The need for human oversight extends beyond simple code verification. Cyber threats, including the recent hacking of supermarket cooling systems and a breach at the cryptocurrency exchange Binance, highlight the limitations of AI agents.
“To prevent this, we need people who can close security gaps,” said Hermans. “That requires a lot of knowledge of how software is put together, and you cannot just ask an AI agent to do that.”
Students like Dolf, a fourth-year data science and AI student, argue that human intervention is necessary to prevent errors. “AI just predicts something, but does not understand if it makes a mistake or not,” Dolf said. “That is why many people are still needed to oversee AI so it does not all end up in a mess.”
Hermans suggests that ICT programs may need to evolve to remain relevant and attractive to new students. She advocates for curricula that emphasize the social dimensions of technology and the “why” behind software architecture.
“I think it is important that computer science programs become less technical and that we take a more social approach as a field,” Hermans said.
“We also have to explain to students: why does this code work this way? Why do people need security updates?”
For many, the daily task of reviewing AI-generated code can be exhausting. Hermans said that providing students with a deeper foundational knowledge could make working with AI more enjoyable again.
