Use this checklist to assess whether your course assessment remains valid in a GenAI-enabled environment. It supports critical evaluation and (re)design of assessments and is intended for lecturer reflection and discussion, not as an evaluation tool for Examinations Boards.

Last updated: May 2026.

Answer each question with ✓ or ✗

A ✗ indicates a potential risk that students could complete the assessment or pass the course without engaging in the intended learning process or achieving the learning objectives due to misuse of AI. The greatest risks occur in unsupervised assignments prepared outside a controlled exam setting: written work, presentations, posters, podcasts and videos.

1. Exposure to unsupervised assignments

Question: Is the final grade (partly) determined by unsupervised assignments? Also include assignments that do not directly count towards the final grade but do contribute to passing the course, for example AVV/NAV, foundational work, or compensation for insufficient grades in other parts of the course.

No unsupervised assessment
No credits (EC) are awarded based on (grades for) assignments that students complete without supervision.Advice: There is a very low risk that students will pass the course through unauthorised use of AI; the other questions do not need to be answered.

Unsupervised assessment contributes to final grade
The final grade is determined by one or more assignments that students complete without supervision.Advice: In unsupervised contexts, assume that students may use GenAI to some extent. To ensure their learning process is not disrupted, either convert the assignment into supervised assessment (on location) or adapt the assignment so students still learn despite using GenAI—this could even be a learning objective. Continue with the remaining questions.

2. AI substitution risk

Question: When you run the assignments through GenAI, refining prompts as a student might (e.g. uploading the assignment description, assessment criteria, rubric, and required reading) and you assess the output: could it lead to a passing grade with or without minor changes? Note: this gives only a rough indication as some students make highly advanced, iterative use of multiple tools.

Output has limited utility
The output is of little to no use, or would not lead to a passing grade in its current form. The student would need to refine the prompts considerably and use additional subject-specific knowledge and/or skills in order to produce a final product of passing quality.

Output could achieve passing grade
The output could lead to a passing grade without any changes, or with minimal changes or refined prompting.Advice: This assignment should not be administered without supervision. Make it an on-site assignment (without the use of aids). If that is not possible, you can increase the complexity of the assignment by using unique, highly specific or recent cases and asking for subject-specific knowledge, skills or personal analysis, whereby students also go through the desired learning process if they do use GenAI. Please note: with advanced use of GenAI, more is often possible than you might think.

3. Weight of unsupervised assessment

Question: Do the unsupervised assignments have a significant impact on the final mark for this course? In other words, could a student pass this course by only scoring well on this component?

Other forms of assessment are primary
The final grade is primarily determined by other forms of assessment, supervised and on location. For example: an exam or assignment, a practical test or an oral exam.

Advice: if the final grade is primarily determined by other forms of assessment, it may be an acceptable risk if students use GenAI for the unsupervised assignment. Check whether the (most important) learning objectives and/or skills are covered by the current assessment or, for example, return later in the curriculum and are assessed in a supervised setting.

Unsupervised assessment dominates
The final grade is based (almost) exclusively on unsupervised assignments.

Advice: convert (part of) the unsupervised assessment to a supervised assessment form as mentioned above. Consider the constructive alignment of the course: can you still assess the same learning objectives/skills this way?

One possible adjustment: have students complete an unsupervised assignment (for AVV/NAV or a small percentage of the final grade), then during a supervised exam, have them do something new with this material (e.g. convert it into a different form or for a different audience).

4. Visibility of the learning process

Question: Can the lecturer observe the student’s learning process? Does the assignment provide sufficient insight into how the student arrived at the final product?

Examples of visible learning processes

Student progress is monitored during the course through:
Draft submissions and feedback
Discussions of research design or argument structure
Parts of the assignment completed in class
Reflective accounts of how AI tools were used
Logbooks documenting the development of the work
Oral explanations or defences of the final product

These elements may also form part of the assessment.

Learning process is visible
Student progress is monitored during the course. For example: students submit drafts for (peer) feedback, discuss their approach with the lecturer, complete part of the work on campus, or when submitting the final product, provide a logbook, oral explanation, defence of the learning process, and/or account of GenAI use. For group work, steps distinguish individual contributions. These components may form part of the assessment to encourage responsible GenAI use (e.g. as a small percentage of the grade or as AVV/NAV).

Learning process is not visible.
Much or all of the work process takes place outside the lecturer’s supervision.Advice: Incorporate scaffolded steps as described above to emphasise the student’s learning process within your teaching. It is not possible to establish with certainty whether the student achieved the learning objectives based on the final product alone.

5. Validity of assessment criteria

Question: Are assessment criteria focused on reasoning rather than polish
Assessment criteria should primarily evaluate students’ reasoning, interpretation, disciplinary judgement, and methodological choices, rather than surface-level features that can easily be generated or improved by GenAI tools. One major risk of GenAI tools is that they can quickly generate highly-polished content without substance.

Criteria focussing on reasoning
Assessment criteria primarily evaluate:

Quality of reasoning and argumentation
Interpretation of evidence or sources
Disciplinary judgement and methodological choices
Originality of analysis or synthesis

These aspects require students to engage with underlying learning objectives, even when using GenAI tools.

Criteria emphasising surface-level features
Assessment criteria place substantial emphasis on features easily generated or improved by GenAI tools, such as:

Grammar, spelling, and language polish
Formatting, layout and structure
Stylistic refinement/surface language use

Advice: Shift the focus of grading criteria towards reasoning, analysis, and disciplinary decision-making. Treat surface-level features as AVV/NAV requirements or submission prerequisites rather than significant grade components. If these aspects correspond to important learning objectives (e.g. language proficiency), assess them under supervised conditions.

6. Transparency and guidance on AI use

Question: Do students know how to use GenAI responsibly in this course? Is the role of GenAI explicitly addressed?

Clear communication about GenAI helps students use these tools in an informed and responsible way.

Students should understand:

When independent mastery is required (and why)
When AI use is permitted
How AI use should be documented (if applicable)
That they remain responsible for all work they submit

Clear guidance on GenAI use
Students know how to use GenAI responsibly in the course. The lecturer explicitly discusses the ethical and academic implications of using GenAI in assignments and encourages students to reflect on the advantages and disadvantages of doing so. For assignments that are (partly) completed without supervision, students are expected to account for any use of GenAI (e.g. through reflection, a logbook or an oral explanation).

Guidance is unclear or use is prohibited
It is not clearly described how students may use GenAI, or its use is explicitly prohibited (even though this cannot be enforced).

Advice: A complete ban on GenAI is difficult to monitor or enforce and may discourage students from discussing their use with their lecturer—even when they have legitimate questions. Clearly explain how GenAI may be used ethically, informedly, and responsibly. Encourage students to reflect on ethical and academic implications, account for use via logbook or reflection.

Interpreting checklist results

If completing the checklist resulted in multiple ✗ answers, it may indicate that parts of the assessment are vulnerable to AI replacing the intended learning process. This does not necessarily mean that the assessment is flawed. However, it may be useful to reconsider certain elements of the assessment design.

There is no one-size-fits-all solution. You may need to combine several strategies depending on the course context. Rules and practices surrounding assessment vary between programmes and faculties, which means that some adjustments may not be feasible in your situation.

Via TLC Contact, you can contact your faculty’s assessment specialists. You can discuss potential changes to your assessment with them. You can also seek advice from the assessment specialists at TLC Central (tlc@uva.nl). Make sure to inform your programme director, so that they can make an overview of the situation and risks at programme level.