Generative AI Enables 'A' Exams with 'Zero' Understanding, Norwegian Educator Warns

The increasing sophistication of generative artificial intelligence presents a significant challenge to educational assessment, allowing students to create impressive, well-articulated texts on subjects they have minimal understanding of. This capability raises concerns that students might pass exams without acquiring genuine knowledge, becoming unable to explain the content if questioned directly.

“

With generative AI, a student can produce impressive, well-articulated texts on topics she barely understands – and become speechless if you ask her to explain.

Høyskolelektor Håkon Njøten highlights that educational research has long emphasized that understanding is built through cognitive resistance. When AI tools remove this resistance, they inadvertently remove the learning process itself. Njøten argues that competence is developed internally within a person, not externally through AI-generated content. This distinction is crucial: one goal is to assess what a student can create using all available tools, including AI, which is a necessary skill in 2026. The other, more fundamental goal is to verify that the student, not the AI, truly understands the material.

“

Knowledge is built in the human, not in the machine.

Njøten dismisses the idea of using reflection notes as a solution, noting that AI can easily generate these as well. He advocates for a more direct approach: oral examinations. A brief conversation about a student's work can reveal their understanding within minutes, especially when conducted without the barrier of a screen. While acknowledging the need for students to learn how to use AI tools effectively, Njøten stresses that this must be balanced with ensuring they possess sufficient knowledge to avoid being replaced by the technology.

“

A short oral conversation about one's own work reveals understanding within minutes.

The article suggests that current assessment methods may be insufficient. An engineer who blindly accepts AI-generated code or an accountant who doesn't understand an AI-prepared financial statement could pose significant risks. Therefore, educational institutions need a dual approach: open assessments that allow students to utilize AI tools, and closed assessments, such as oral exams, to confirm individual comprehension and judgment. The ultimate goal is to ensure that academic credentials signify actual knowledge and critical thinking, not just proficiency in using AI.

“

We need both open assessments where students create with all available tools, and closed assessments where we ask about the individual's own understanding and judgment.