The AI Compass at HVL – a guide to using generative AI in learning and assessment

This guidance is for module coordinators, teaching staff and students. It explains when and how students may use generative AI in assessments in a sound academic and ethical way. The goal is to promote reflective and transparent practice that supports both learning and assessment. The guidance is pedagogical and practical in focus; it is not a legal document.

Important to know

The Regulations on Studies and Examinations at HVL and HVL’s Guidelines on Academic Misconduct apply to all submissions, regardless of the AI Compass.

What do we mean by generative AI?

Generative AI refers to tools that can generate, for example, text, images, music or code. See the overview of recommended AI tools for staff and students at HVL here.

For module coordinators and teaching staff: how to use the AI Compass

  • Decide the permitted level of AI use for each compulsory learning activity (Norwegian: OLA, “obligatorisk læringsaktivitet”) and for the exam.
  • State the level and the required documentation in the assignment brief.
  • Provide students with written information in advance about:
    • which level applies
    • how they must document any AI use
  • Adjust the scope of the assignment if documentation/reflection on AI use will be part of the submission. Specify if documentation should be included as an appendix.
  • Rules for citing and referencing AI tools may vary between referencing styles. The Norwegian APA Manual v.1.12 allows local adaptation. Make it clear which rules apply in your module.

Three levels of AI use

The module coordinator/teacher sets the level of AI use for each OLA and for the exam. You may tailor the description of the level to each assignment.

1) Exploratory use of generative AI

Use this level when you want the student to work actively with generative AI throughout the process.

The student: 

  • may use AI tools creatively and critically for ideas, text, images, code, analysis, etc.
  • may use AI to carry out analysis if the entire process is documented
  • is responsible for the content and must evaluate and stand by the results
  • must clearly mark any content taken directly from an AI tool (e.g., with quotation marks, italics or highlighting)
  • must briefly reflect in the text on how AI use affects the content, academic quality, their own learning and working methods

Documentation

  • In the methods section or an appendix, the student must state which tools were used, how they were used and why.
  • If AI use has influenced results, analysis or findings, the tool may be cited in the text and included in the reference list in line with the module’s referencing style.

2) Generative AI as a support tool

Use this level when the student may use AI as support during the work, but with clearly defined limitations.

The student:

  • may use AI for language polishing, structure, outlining and idea generation
  • must not copy and paste text directly from AI; all content must be reworked and integrated with their own reasoning
  • is fully responsible for the content of the submitted work

Documentation:

  • In the methods section, introduction, or a separate section before the reference list, the student must state which tools were used, how they were used and why this use was appropriate.
  • If AI was not used, the student may include a short note explaining why.

3) Assessment without the use of generative AI

This level is best suited to invigilated exams (on‑campus written exams, oral exams or practical exams). For take‑home exams, portfolio assessment and unsupervised assignments, it is difficult to ensure that the work has been completed without AI; this level is therefore less suitable for such formats.

Documentation:

  • None

 Tips for students: how to document AI use

Placement:

  • If AI is used in analysis or in presenting results (figures, tables, numbers, etc.): document AI use within the assignment where it naturally fits, for example in the methods section.
  • If AI is used for language, structure or outlining: include a separate section outside the main body of the assignment, for example immediately before the reference list or as an appendix.

Content:

  • AI tool (name and, if applicable, version), date and purpose.
  • How you used the tool (a brief description of the steps).
  • Your evaluation of the quality and limitations of the AI output.
  • How you checked, edited and validated the results.
  • Any key prompts (if relevant to understanding your process).

Citations and references:

  • Follow the module’s referencing style and guidance (e.g., Norwegian APA Manual v.1.12).

 Example formulations for documentation

These formulations are based on Levels 1 and 2. Adapt them to your assignment.

Level 1 – exploratory use:

  • “The code/algorithms in this assignment were developed with assistance from [tool]. I have verified, tested and take full responsibility for the results.”
  • “[Tool] was used to produce figures, tables, numerical values and similar. I have ensured that the presentation is consistent with my data.”
  • “One or more images/figures were generated using [tool]. These are clearly labelled.”
  • Other types of generative AI use: specify briefly.

Level 2- AI as a support tool:

  • “I wrote the first draft myself and used [tool] for language polishing and structure. I have reworked any AI suggestions and take full responsibility for the content.”
  • “Parts of the text were revised and structured with support from [tool]. All content has been checked and quality‑assured by me.”
  • “Some ideas or approaches were suggested via [tool]. I explain how I used them in the assignment.”

 Resources

  • The “AI Compass” is an adapted version of the “Framework for Assessment and AI (VuKI)” at Østfold University College (HiØ), based on the “AI Assessment Scale (AIAS)” (Perkins et al., 2024).
  • National Digital Learning Arena (NDLA): “Chatbots and assessment: the VuKI model” (Nynorsk).
  • Sources and AI use in academic work: Søk & Skriv.

 About this guidance

The AI Compass was developed by the AI Coordination Group at HVL (Western Norway University of Applied Sciences). The first version was released in September 2025 and was called “the Traffic Light Model” (“Trafikklysmodellen”). This revised edition, “AI Compass”, was approved by HVL’s Education Committee (“Utdanningsutvalet”) on 26 March 2026 and published on hvl.no on 30 April 2026. In this work we used an AI service from OpenAI and the plain‑language tool “Klarspråkshjelpen” from NAV.