GenAI and assessment at course level

 

1. Is the final course grade determined (in part) by written assignments that students complete without supervision?

 

	No credits (EC) are awarded based on grades for assignments that students complete without supervision. There is very little risk that students may pass the course through unauthorized use of GenAI; there is no need to answer the other questions.
	Yes, the course grade is determined by one or more assignments that students complete without supervision. Continue with the other questions.

2. How much of the final grade is determined by unsupervised assignments?

 

	The final grade is primarily determined by other forms of assessment where there is a lower risk of plagiarism and cheating (for instance an exam on location, a practical test, or an oral exam).
	The final course grade is determined almost exclusively (or even completely) by unsupervised assignments.

3. Do the assignments allow the lecturer to monitor the student’s progression towards the final product?

	The lecturer monitors the student’s progression during the course, for instance, by having students hand in drafts for (peer) feedback, discuss their approach with the lecturer, complete part of the work on campus (in class), or provide an (oral) account of the work process.
	The whole work process takes place outside the lecturer’s supervision.

4. Do the assignments allow students to pass without course-specific knowledge or skills?

 

	The assignments require a subject-specific level of depth and rely less on generally available knowledge. In terms of content, format and/or method, the assignments are specifically attuned to what was learned in the course and during the classes. (NB: GenAI is getting better at highly specific tasks, using the right prompting.)
	Students can make their own choices regarding content (theme), format and/or method, using generally available knowledge.

5. Run the assignments through GenAI, refining the prompting as a student might. Evaluate the output – could it lead to a passing result, with or without minor changes?

 

	The output is of little to no use, or it would not lead to a passing result in its current form. The student would have to refine the prompting considerably as well as use additional subject-specific knowledge and/or skills.
	The output would lead to a passing result with no changes, or it could lead to a passing result with minimal changes or refined prompting.

6. Is there a clear policy on student use of GenAI in the course?

	Students know whether/how they may use GenAI responsibly in the course. The lecturer explicitly discusses the implications of GenAI use in assignments.
	There is no clear policy on this (remember that banning the use of GenAI altogether only works on paper as there is no way of checking this: detection tools are not reliable).

The detection tools that are currently available are not reliable and should not be used. They cannot accurately predict whether a text was written (in part) by GenAI or a human.

Unsupervised (written) assignments are highly vulnerable to GenAI misuse. See here for more information [Link to page GenAI and written assignments]. Go back to your learning outcomes. Is this the only form of assessment you can use to determine whether students have met the learning outcomes? Can the weighting of the written assignments be reduced in favor of other types of assessments? Are there other ways of assessing the same knowledge/skills? Take a look here for inspiration: 1. Designing – UvA Teaching and Learning Centres (TLC).

The most common tips for preventing unauthorized use of GenAI in assessment are:

• Monitor students’ progression towards the final product (formative assessment). Have them hand in outlines or drafts, meet with you to discuss their approach, document their work process, or even complete part of the work on location. Make sure your setup is feasible within the allocated hours for the course.
• Have students complete assignments or take exams on location, especially for lower-level cognitive skills such as remembering, understanding, and applying. This approach may also work well if you want to assess foundational academic writing skills such as summarizing and paraphrasing. It is less suitable for assessing higher-level cognitive skills such as analyzing, evaluating, and creating. Do note that there is a shortage of exam locations: talk to your programme coordinator about this.
• Use oral assessments if this is feasible in terms of time and staffing. Oral assessment can take place online or offline, but always in real time. There are a variety of types of oral assessments you can use: oral exam, orally explaining a written assignment (all students or a random sample), presentations for all students (e.g. a pitch), poster presentations, group discussions, debates, roleplaying, simulations/demonstrations (such as patient-doctor interaction, a negotiation simulation, moot court, business or code demonstrations), Q&A session by students, interviews.
• Use assignments that require creativity and originality, such as blogs, vlogs, video essays, music videos, poetry, infographics, animations, podcasts, portfolios, website design, etc. Check that any skills you assess are taught in your class or elsewhere in the degree. In other words, make sure students aren’t penalized for not being good at skills such as coding, graphic design, or video editing if these are not part of the learning outcomes.

Especially if you are using unsupervised assignments, make sure that the final grade is also based on other forms of supervised assessment without access to GenAI, so you can be certain that students have actually met the learning outcomes.

The more widely available the knowledge and skills you are assessing, the easier it is for students to use GenAI to do the work for them. Do not assess such learning outcomes without supervision. If you are using unsupervised assignments, make sure they are hard to complete with help from GenAI by requiring students to integrate (personal) examples, case studies, theories, or research that is highly specific to the course.

Help students understand plagiarism and fraud and evaluate GenAI output. Discuss the implications of using GenAI for assignments and explain the value of learning how to do research yourself. It also helps to appeal to students’ motivation, explaining the value of the assignment you’re asking them to complete: What are the learning outcomes? How will this help them in the long term? What are the benefits of learning to do this without help from GenAI?

Provide a clear policy on GenAI use for your course or assignment so that students know whether and how they can use GenAI responsibly and ethically.