While the use of GenAI by students may sometimes not support their learning, it can also be a valuable tool that helps students develop new skills and approaches in our discipline. Additionally, these tools can be leveraged to create more personalized and engaging learning experiences.
We hear from our students that they use GenAI when they are feeling stressed and anxious about their grades.
Include In-class Elements
Bringing assessment or elements of assessments into the classroom can help reduce the risk of students using Generative AI in their submissions.
- In-person exams, such as midterms and final exams limit AI misuse.
- In-person presentations and interviews allow for real-time assessment reducing AI use.
- Including in-person elements as part of larger assignments ensures active participation and reduces AI-driven submissions.
- Graded group work, peer-review, Q&As, debates, and other active learning approaches foster collaboration and limit AI usage.
- Case-based and problem-based learning engage students in real-world scenarios, and minimize AI-generated responses.
Emphasize Process rather than Product
Making the process more visible helps ensure the authenticity of student work and reduces the likelihood of AI misuse.
- Deemphasizing the weight of the final product by including multiple steps such as outlining, drafting, peer-reviewing, and accepting/rejecting feedback emphasizes the process over the outcome.
- Providing ways for students to explain how they came up with a result or answer ensures transparency and reduces reliance on AI.
- When working in teams, students co-create a team contract to promote accountability and minimize the use of AI-generated work.
- Post-submission interviews allow instructors to verify the student’s understanding and reduce AI-related concerns.
Design Multimodal Assignments
Utilizing various formats for assignments enhances engagement and reduces reliance on generative AI, as current AI tools do not effectively generate multimodal outputs.
- In addition to traditional research paper-style writing, students may demonstrate their learning through videos, podcasts, infographics, etc.
- Students annotate videos or other media and are evaluated based on their annotations, ensuring active engagement and reducing AI dependency.
Emphasize Tasks That AI Cannot Perform Well
Educators have been developing alternative assessments that ChatGPT cannot easily complete. This approach carries risks, given the rapid evolution of technology and the continuous addition of massive amounts of prompts, which improve the outputs generated. However, while making assessments completely AI-proof may be nearly impossible, we can focus on tasks in assessment expectations that foster personalized and localized connections. For example, demonstrating learning could involve linking to local contexts or current events, which may not be well represented in large language models.
Update Grading Schemes
You might consider updating your current grading schemes, rubrics, and outlined criteria to prioritize and emphasize competencies that are less mechanical and harder for AI to replicate. For instance, in a writing assignment, you could shift the focus towards skills like constructing solid arguments, incorporating and assessing evidence, and conducting critical analysis, rather than emphasizing grammar and essay structure. Additionally, asking students to submit rough planning notes with their assignments could help evaluate their process, not just the final product.