Generative AI in the Exam Room — Denmark's VIVE-Evaluated Afsætning A Trial at HHX Business Schools

Denmark's Agency for Education and Quality (Styrelsen for Undervisning og Kvalitet) has been running a multi-year trial of a redesigned exam format for Afsætning A (Marketing A), a subject on the business-focused upper-secondary track (HHX), that explicitly permits students to use generative AI as part of the assessment rather than banning it. Ten schools took part in the summer 2025 trial exam, combining written, oral and group-work elements with sanctioned generative-AI use, alongside two related trial formats also covered by the same evaluation.

VIVE, Denmark's national social and education research institute, conducted the independent evaluation, published in January 2026. It found that the average grade across the trial exams was at least one full grade point higher than the average for the ordinary written and oral Afsætning A exam in both 2023 and 2024 at the same institutions -- a consistent roughly one-point increase across all three trial formats compared with the prior year's ordinary exams. Participating teachers judged that the new format better assessed genuine student competencies, but VIVE was explicit that its dataset could not establish why grades rose: whether from genuinely stronger demonstrated competency, a shift in what the redesigned format rewards, or other design effects specific to the trial cohort and schools was left as an unresolved question rather than assumed to be the former.

Based on the 2025 evaluation, the Agency decided to continue the generative-AI trial into 2026 and 2027, moving from locally designed exam cases to a centrally designed case used either as a group or an individual exam, with all second-year HHX classes eligible to register. The programme sits alongside a separate, more restrictive Danish trial in English A oral exams (60 minutes' preparation with AI tools permitted, followed by an in-person oral exam without them), evaluated separately by the Ministry of Children and Education, reflecting a broader national effort to test several distinct AI-and-assessment formats in parallel rather than adopt one national policy outright.

Read the full analysis: https://www.vive.dk/da/udgivelser/evaluering-af-proeveforsoeg-jvnm5ryx/