Discussion

Norway’s AI report urges universities to rethink exams and build staff expertise

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#13501

Norway’s committee on AI in higher education has presented recommendations that include more supervised exams, fewer short take-home exams and better AI training for university staff. Its report also raises a practical challenge for universities: written work alone can no longer reliably show what a student knows.

Watch Desk analysis

What happened

The committee, appointed by the Norwegian government in March 2025, presented its findings to Research and Higher Education Minister Eileen Fugelsnes last week. Science Norway reports that the recommendations include giving lecturers sufficient expertise to guide students’ responsible use of AI, and exploring AI systems that explain grading decisions or help students understand their academic performance.

For students, the committee offers nine recommendations intended to support learning rather than let AI do all the thinking. It advises against taking shortcuts and recommends that universities make greater use of supervised, in-person examinations while phasing out short home exams.

Why it matters

The proposals go beyond deciding whether students may use AI. They ask universities to revisit what counts as evidence of learning, how assessments should work and whether staff have the expertise to guide students. The committee says AI may harm learning when appropriate teaching measures are absent, and recommends institutions keep assessing how available systems affect different subjects and professional practice.

The report also leaves room for local decisions. Researcher Christine Sætre tells Science Norway that an overarching framework must be put into practice in each institution, and that AI-supported marking needs further research with people responsible for quality assurance.

Our read

This is a useful shift from trying to catch AI in student work towards asking what an assessment is meant to measure. More supervised exams may help, but they are not a complete answer. Universities will need to pair clear rules with teaching that helps students use AI without outsourcing the learning itself. A policy document is a start; the timetable and the staff training are where the hard work begins.

What to watch

  • How universities translate the committee’s recommendations into assessment rules.
  • Whether staff receive practical AI training, rather than guidance that stays on paper.
  • What research finds about AI-assisted grading and its explanations.
  • Whether institutions change assessment formats while preserving fair ways to demonstrate learning.

Discussion spark: Should universities respond to generative AI by bringing back more supervised exams, or redesigning assessment so students can use AI openly while still showing what they know?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.