Strategies for reducing reliance on GenAI in assessment
Find strategies for mitigating students' reliance on GenAI and promoting human-centred learning outcomes.
Keeping students from relying on GenAI is important from the standpoint of academic integrity: students need to own and take responsibility for their learning, and instructors need to know that students’ work is their own. But strategies intended to keep students from relying on GenAI aren’t just about keeping students from engaging in academic misconduct, they can in of themselves be good for learning.
Shifting some assessments to in-class, valuing process alongside product and encouraging students to practice skills that GenAI is not good at (at least for now) can mitigate reliance on GenAI and allow students to achieve human-centred learning outcomes.
In-class assessments
Students have opportunities to complete in-class verbal or written exercises, where they create and share their ideas with other students during class time. To reduce stress on these types of high-stakes assignments, consider the following adjustments to in-class assessments:
- Allowing a “cheat sheet” or similar aid
- Making it an “open-book” exam
- Providing questions in advance
- Requiring only the first draft to be written in class
Process over product
In process-oriented approaches, students complete components of a paper or project in the way of an outline, drafts, and/or annotated bibliographies before submitting a final version. Students can also be asked to document the differences across drafts of a paper, highlighting what they did and the effect it had on their work.
This approach can work for other types of final submissions beyond papers or essays:
- A portfolio can be preceded by two or three submissions of sample artifacts alongside students’ reflections, a proposal for resolving a problem might be scaffolded through a problem statement, a solution proposal and a cost/benefit analysis.
- For courses in which exams are the principle means of assessment, a two-stage exam wherein students practice collaboration and divide the testing process in two may be a good solution.
Emphasize tasks GenAI can’t do (or can’t do well – for now)
- Ask students to submit pre- and/or post-assignment reflections in writing and then follow up verbally as a basis for an in-class discussion.
- Design assessments that have students identifying “tasks that require discernment, call for an explanation, compel reflection, and ask questions like ‘why’ and ‘should.’” (Jenner 2025).
- Encourage human collaboration by assigning group work, peer-review, Q&As, and debates.
- Invite students to relate discussion posts, reflections, short assignments, etc. specifically to discussions from the lecture or in-class activity (assuming that information is not on slides shared with students)
- Make assessment tasks more meaningful for students by inviting them to personalize them and include their lived experience or develop an assignment around experiential learning such as a field visit, an invited lecturer, role play, case studies, real-world challenges.
Disclosure notice: This resource was developed and written by a human without Generative AI assistance and was revised based on peer feedback. Microsoft Copilot was used in the formatting of the references, and its accuracy was checked.