I want to prohibit AI in my classes:

Creating AI-Resistant Assignments

AI-Resistant assignments are assignments that attempt to mitigate the use of generative AI (genAI) by students. Some of these strategies may be intuitive, and some you may already have implemented.

General Strategies and Best Practices

Set Expectations Early

Most students want to be academically honest and are hesitant to use tools that may be considered cheating (check out our student survey data for a deep dive). The problem arises when students: a) do not have clear expectations of acceptable use, and b) do not have enough AI or digital literacy to understand which tools are genAI and which are not. Instructors can increase student confidence by providing clear expectations for acceptable genAI use in their syllabus and reinforcing these expectations in assignment instructions. The new syllabus template provides some sample AI policy language to get you started. There are further examples in the resources section.
Embrace personalization

GenAI is quite good at creating surface-level, general answers and solutions. However, it struggles with replicating personal, local, or unique answers. Instructors can use this when crafting assignments.
1. 1. Connect your assignments to personal/hyperlocal issues: Have your student choose a topic that is personal to them, or hyper-local to your community, campus, or class.
  2. Use metacognition: Metacognition is the awareness and understanding of one's own thought process. Students reflecting on their own process or synthesizing their insights on assignments is hard to fake with AI, and also encourages deeper engagement with the material.
Consider Assessments as Learning Experiences

Any time instructors can incorporate experiential or scaffolded learning, their assignments will be more resistant to genAI and promote active learning.
1. 1. Build Scaffolded Assignments: As mentioned above, any assignments that are scaffolded make it hard to use genAI. Assignments such as portfolios, turning in outlines or drafts, and in-class revisions hold students accountable while also allowing them to see and reflect on their own growth.
  2. Adopt Non-Traditional Assessments: Could student learning be assessed through a performance, debate, or simulation? The very nature of these types of assessments does not allow for genAI use. However, instructors should also have students turn in their prep or process documents to further mitigate genAI usage. Instructors have access to BodySwap, a simulation tool.
  3. Use Project-Based Assessments: Is there a real-world problem or challenge that students could tackle? Creating tangible solutions or products tends to be more engaging and impactful with students.
Encourage Collaborative Learning

Using genAI is difficult in a group learning context. Students are more likely to call out genAI use amongst their peers. Collaborative assignments/assessments also build communication, teamwork, and problem-solving skills.
1. 1. Include Students in Building Assessment Outcomes: Including students when building assessment outcomes and rubrics encourages student buy-in and gives students clearer expectations of deliverables.
  2. Consider Collaborative Annotation: Use tools like Perusall, Hypothes.is, or Harmonize for students to collaboratively annotate readings or videos. Collaborate (on the left sidebar) works for assigning group papers or notes. This has the added bonus of making students actively engage with class materials.
  3. Incorporate Peer Review or Live Discussions: Collaborative and peer-review assignments introduce students to diverse perspectives and new ways of thinking. KCC has several tools to facilitate collaboration through Canvas: Harmonize makes peer reviews easy by automating the assignment of peer reviewers, generating rubrics, and assigning multiple due dates for initial and response posting. The Big Blue Button and Lucid allow for real-time brainstorming, discussions, or collaborative mind mapping.
Test Assignments with GenAI

Before finalizing assignments, have a genAI attempt them. This will allow instructors to see which parts of the assignment are easily completed by AI. Remove or alter those parts so that the assignment assesses your learning outcomes and not rote retrieval (unless that is one of your outcomes).
Go Analog

If all else fails, return to the tried and true methods of assessment. Reintroduce pen-and-paper assignments, in-class essays, scantrons, and proctored exams.

Specific Discipline AI-Resistant Assignments

This page was created with grammar and spelling corrections by Grammarly. The content on this page is a synthesis of guidance from the NMU CTL, The University of Chicago, The Teaching Channel, CT State Community College, NIU Center for Innovative Teaching and Learning, and the Washington University in St. Louis CTL.