How AI is changing work
Education
Learning with AI, and learning what AI cannot do.
Pupils and students already use generative AI widely, teachers more cautiously. Schools and universities are making agreements, building their own safe AI environments and adapting their assessment. The question is shifting from what a student produces to how it came about.
What changes
- Lesson preparation: teachers have AI draft lesson plans and practice material and then adapt them.
- Assessment: more oral explanation, interim products and accountability for AI use.
- Feedback: AI tutors and chatbots give students step-by-step explanations and feedback.
- Own infrastructure: institutions offer AI chat in which data stays in-house.
Where people make the difference
- The relationship between teacher and pupil, and seeing what a child needs.
- Judging whether someone really understands something, and the conversation about it.
- Choices about what good education is: which skills a pupil must master themselves.
How AI is already used
School agreements about AI
Kennisnet, together with the VO-raad, the PO-Raad and the Ministry of Education (OCW), made a toolkit that helps schools make agreements about AI, with worksheets, a letter to parents and seven principles for responsible use.
VO-raadThe university's own AI chat
Since September 2025, the University of Amsterdam has offered its own AI chat to students and lecturers. What is shared in it stays within the UvA's infrastructure.
UvAA national AI environment
SURF and Npuls are building an AI hub and chat application for all of vocational, applied and academic higher education. Institutions choose which models they make available.
SURFAgents that search sources reliably
Rotterdam University of Applied Sciences is using SURF's AI hub to build agents that search scientific databases with fewer made-up sources.
SURFAccounting for how you used AI
According to 70% of examination boards in higher education, students must (soon) explain what instructions they gave AI or how they checked the outcome.
Inspectie van het OnderwijsRedesigning assessment
Utrecht University advises programmes to choose per learning outcome: assess without AI using interim products and oral discussion, or deliberately allow AI.
Universiteit UtrechtAI tutors, still on trial
Kennisnet describes AI tutors that lead pupils step by step towards understanding instead of giving the answer. The effects on learning have hardly been researched yet.
Kennisnet
Where the sector is heading
A majority of students at universities of applied sciences and research universities use AI at least weekly for their studies (75% and 69%). A large share of teachers feel they do not yet have the knowledge to use AI well (88% in primary, 80% in secondary education).
Dutch Inspectorate of Education, State of Education 2026From January 2027, vocational schools (mbo), universities of applied sciences and research universities will get access through SURF to their own AI infrastructure with open and commercial language models.
SURFAI literacy among education professionals is becoming a topic in its own right, with a roadmap for primary schools (March 2026).
Arbeidsmarktplatform PO
What a maker can do for you here
- “Build a practice assistant that helps pupils without giving them the answer.”
- “Make teaching materials and explainer videos at different levels to go with our textbook series.”
- “Give our teachers practical training in working with AI.”
- “Set up an AI chat that uses our own study guide and regulations as its source.”
Researched on 5 October 2026. Know a good example from the Netherlands? Let us know.