GenAI-Resilient Assignments

Generative AI (GenAI) tools present exciting opportunities for teaching and learning, but also raise serious concerns about rigor and academic integrity. In a GenAI-saturated environment, academic integrity must extend beyond detecting misuse. Educators can foster integrity by designing learning experiences that encourage genuine engagement, emphasize the need for students to develop their own voice and critical thinking skills, and make shortcuts less useful or appealing. 

Whether you allow, restrict, or integrate GenAI in your classroom, thoughtful assignment design is our best strategy for promoting meaningful learning and upholding academic integrity. Beyond the individual strategies outlined here, a programmatic approach can support students in developing GenAI use as a cumulative skill set throughout their college career, with discipline-specific frameworks and high-integrity rubrics that emphasize GenAI literacy. 

Why do Students Use GenAI?

Based on findings from UMN’s Student Experience in the Research University survey and student focus groups (SERU, DeLorme), common reasons students use GenAI include:

  • Unclear or vague assignment instructions
  • Feeling overwhelmed, behind, or disconnected
  • Low motivation in courses outside of their major and/or areas of interest
  • Understanding of an assignment or task as busywork
  • Insecurity about their writing or communication skills
  • Perceived approachability, skill, interest, and engagement of the instructor 

In response, instructors can:

  • Be transparent about your GenAI course policy and connect assignments to larger course, discipline, or industry goals.
  • Design your assignments so that they have a clear relevance and connection for students.
  • Clearly articulate why students need to be able to write and think critically without GenAI.
  • Reinforce the inaccuracies, biases, and other weaknesses of GenAI.
  • Provide scaffolding and feedback that reduces the need to “sound smarter” through GenAI.
  • Develop shared departmental language around GenAI policies so students become familiar with your discipline’s norms and expectations.
  • Talk to students about the relationship between GenAI and academic honesty.

What Tools Can Inform Your Teaching?

You don’t have to become a GenAI expert, but experimenting with GenAI tools can help you design assignments that more effectively respond to their use. 

Try this:

Spend 30 to 60 minutes using one of the UMN-licensed GenAI tools to produce a response to one of your assignments. 

  • What did the tool get right? Where did it fall short?
  • Would you accept this as a student submission?
  • What parts of the assignment required true understanding, synthesis, or disciplinary knowledge?
  • How would you explain to students the pitfalls and problems with using GenAI to complete this assignment?

When to be GenAI Resilient or GenAI Inclusive?

Designing assignments around GenAI extends beyond simply allowing or banning it in your course. Students need to develop the ability to examine when using GenAI might be valuable, and when it might hinder their own creativity, produce inaccuracies, or create fallacious arguments. 

Consider when you might want to resist GenAI use in your course and when you may want to include GenAI to support your learning outcomes.  Think strategically about your course goals, assessments, and pedagogical strengths: What do you want students to learn? How might GenAI help or hinder those goals? Why is it important to you to have your students articulate their own ideas in their own words? Articulate those rationales in your syllabus and in discussions with students about assignments. 

GenAI Resilient: Design assignments that GenAI tools can’t adequately complete by incorporating real-world relevance, specificity, and diverse modes of expression. This approach focuses on iterative, process-based work, rooted in personal contexts (e.g., lived experiences and opinions, live discussions, group work, in-class presentations). 

GenAI Integrative: When appropriate, GenAI can be incorporated to scaffold learning through critical analysis of outputs and reflective use. This stance incorporates GenAI literacy into the work of the course, asking students to build on GenAI outputs critically and document their use ethically.

Strategies for GenAI-Resilient Assignment Design

Design for Original Thought and Metacognition

Move beyond outcomes that GenAI can achieve with little human engagement, such as rote memorization, formatting, and grammatical correctness. Instead, focus on outcomes that require human skill, such as accuracy, interpretation, creativity, critical analysis, and metacognitive tasks. 

What to do:

  • Incorporate reflection based on specific learning topics from your class: “What did I learn, and how did I learn it?”
  • Teach students to vet and critique GenAI outputs, reminding students that GenAI tools can produce confident-sounding but inaccurate information.
  • Look for ways to promote self-efficacy and explain why it is an important skill: “How do I know I can do this without GenAI?,” “What can I do better than GenAI?,” or “When might I be in a situation where I cannot rely on GenAI?”
  • Align your course assessments with broader program learning goals, making assignments feel more relevant and skills-based rather than task-based. Explain why it is important for students to master those goals and skills independently of GenAI.
  • Studies have shown the value of persistence and challenge in learning (e.g., Deslauriers et al., 2019). Discuss these findings with students and have them reflect on how instances of struggle may contribute to greater outcomes.
  • Develop student-led workshops or exhibitions that require students to engage in teamwork to pool their creativity, share ideas, and come to a consensus (all tasks that GenAI does not do).  Stress that students will likely be expected to carry out these tasks in future employment. 

 

Make Content Personal and Context-Specific

The easiest way to make assignments more GenAI-resilient is to design tasks that GenAI can’t easily complete because they are specific, personal, or rooted in lived experience. GenAI lacks this contextual insight.  

What to do:

  • Tie assignments to specific course content, such as discussions, guest speakers, presentations, events, or disciplinary debates.
  • Ask students to draw on their lived experiences, cultural backgrounds, individual interpretations, or local and current events.
  • Require students to generate their own case studies, project data, or observations.
  • Integrate community-engaged work, service learning, experiential learning, interviews, and/or scenarios.

Focus on Process Over Product

Generative AI excels at producing products that appear polished. Counter this by making students’ work more visible throughout the creation process. 

What to do:

  • Break down large assignments into smaller parts: proposal → draft → revision → final.
  • Require students to submit artifacts of learning: brainstorms, notes, annotated drafts.
  • Provide peer review or instructor feedback opportunities between drafts, encouraging students to revise in response to specific feedback.
  • Make revision a graded component of projects and essays.
  • Have students produce some of their writing during class.
  • Use tools such as Google Docs’ version history to review when and how a document has changed over time.
  • Provide frequent opportunities for written reflections where students describe what they did, how they approached the work, and where they struggled.

Diversify Deliverables

Though GenAI tools are emerging that can produce a range of media, multimodal learning can reduce the likelihood of GenAI misuse and encourage deeper student engagement with course concepts. 

What to do: 

  • Develop assignments that involve performance, discussion, or visual communication, which are harder to outsource to GenAI.
  • Use social annotation tools to conduct a discussion in the margins of a text.
  • Incorporate in-class assessments, handwritten where possible (e.g., timed writing, oral exams, debates).
  • Pair assignments with oral defenses, discussions, or collaborative work.
  • Check understanding with low-stakes assessments such as quizzes, brainstorming, or concept maps.

Grading, Evaluation, and Assessment

Assessments should reward original thinking and give clear incentives for integrity. Ask yourself: Does this assessment tool measure student learning and not just the final product? 

What to do:

  • Create rubrics that evaluate process, originality, accuracy, analysis, evidence of learning, and application of course material.
  • When GenAI is allowed, include assessments for information literacy and ethical use.
  • Provide feedback that focuses on the student’s ideas and growth, not just correctness.
  • Integrate digital tools that support rich, efficient feedback and dialogue, such as audio comments or feedback banks tailored to recurring issues.
  • Incorporate metacognition in your grading or evaluation process to provide students with opportunities to reflect on their learning processes and on how their use of technology impacts their learning. 

Maintain Accessibility, Inclusion, and Equity

As course assignments become more complex, stay attuned to the ways that diverse learners encounter your course. 

Focus on providing: 

  • Clarity: Provide instructions in multiple formats (written, visual, verbal) and examples of successful student work whenever possible.
  • Equity: Avoid assumptions about access to GenAI tools, cultural fluency, or prior academic experiences. Connect all students with academic and technology support resources in your syllabus and on your Canvas site.
  • Flexibility: Make space in the classroom and in assessments for diverse ways of knowing and expressing knowledge.
  • Digital Accessibility: Opt for University-supported academic technology tools that have been vetted for digital accessibility. Apply the 7 core accessibility skills when creating content. 

References and Resources