Business Case August 2026 UiT The Arctic University of Norway
AI Assisted Feedback: Findings from a second, formative pilot at UiT
What changed when the feedback arrived before the exam instead of after it.
This is relevant to you because:
- You want feedback to reach students while it can still change how they prepare.
- You need workload figures before arguing that AI-assisted feedback is viable at cohort scale.
- You require every AI-generated comment to be screened by a human before students see it.
- You want to know how much prompt refinement improved on an earlier pilot's output.
Abstract
This second UiT pilot answers the timing critique raised by the first. On 6 May 2026, 23 medical students sat a formative immunology test in MED-1501 that did not count towards their grade. AI-assisted feedback was generated for every submission and returned four days later, during the teaching period rather than after the exam.
Fourteen students completed the evaluation. Thirteen agreed the feedback arrived at a useful point for learning, eleven found it relevant to a large extent, and all said it helped them understand what to improve. The assessor applied one refined formative prompt across every submission, screened each response before release, and estimated five minutes per student against a twenty-minute manual baseline. The sample is self-selected, so figures are indicative.