Creating AI-Resilient Assignments Workshop

Goals of the workshop

• Define what a process for creating AI-resilient assignments can be.

• Suggest best practices and guidelines for this work.

• Give advice that is practical, specific, and rooting in teaching in Lehigh classrooms.

• Prompts: How do you design assignments and courses given the prevalence of generative AI? What is an AI-resilient assignment? What are some negative and positive implications of your students’ use of AI?

Prompts for panelists

• How do you design assignments and courses given the prevalence of generative AI?

• What is an AI-resilient assignment?

• What are some negative and positive implications of your students’ use of AI?

Towards a definition of “AI-resilient assignments”

• Resilience means assignments that adapt to technological change without losing their core learning objectives.

• An AI-resilient assignment emphasizes process, student ownership, creativity, critical thinking, and the motivation of having a public/ readership for student work, among other aspects.

Assignment design strategies

• Break down assignments into scaffolded, multi-step tasks to emphasize process and reduce opportunities for AI misuse.

• Personalized and reflective writing, such as autobiographical or student-designed assignments, is harder for AI to replicate.

• Public presentations or peer-reviewed work incentivize students to invest effort and originality.

Classroom approaches to AI

• Encourage open conversations about AI use. Talk about if, how, when, and why students use it.

• Teach students to critically assess AI tools and recognize their limitations (e.g., bias, lack of empathy, factual errors).

• Hands-on AI experimentation (e.g., comparing student work with AI-generated versions) helps demystify GenAI and build confidence.

Ethical and pedagogical concerns

• AI detection tools are unreliable and often inequitable (false positives, especially among ESL or neurodiverse students).

• Overreliance on automated AI detection creates a punitive atmosphere. Instead, focus on thoughtful assignment design and communication.

Grading and assessment innovations

• Labor-based or contract grading (as shared by Dr. Suzanne Edwards) shifts emphasis from outcome to effort, growth, and process.

• Contract grading can also reduce pressure that often leads students to misuse AI and promote positive risk-taking and self-directed learning.

Discipline-specific approaches

• AI's role and impact differ across disciplines (e.g., in the particulars panelists discussed related to journalism, religion, engineering, and writing).

• Faculty are encouraged to adapt their approaches to fit the learning goals of their domain and student needs.

Summary of insights from each panelist

• Dr. Haiyan Jia

  • Remarked that students often turn to AI because of stress, time constraints, and feeling hopeless or adrift in a class.

  • Introduced a range of key terms including AI-resistant, AI-critical, and AI-organic.

  • Instructor techniques to make their assignments AI-resilient or AI-resistant include allocating time for project work in the classroom, introducing process-oriented assignments, and assigning deliverables at several points in the semester.

  • It is important to teach students to be AI-critical. This means teaching them media literacy as it pertains to AI, including doing comparative assignments in class (e.g., they do an assignment, then they do the assignment using AI as a teammate/ assistant).

  • To be AI-organic means encouraging computer-human interactions and incorporating AI into our classes in a purposeful way.

  • Her goal is to teach AI critically and organically in journalism. She emphasizes hands-on comparison and interdisciplinary learning.

• Dr. Annabella Pitkin

  • There are benefits to having students cite and reflect on their use of AI as part of an assignment (e.g., within an appendix or addendum to assignments).

  • When students know a topic well, they can easily critique outputs from AI. When they don’t know a subject well, they are more vulnerable to the mistakes and biases embedded in AI.

  • She is committed to reflective AI use through student journaling.

  • A sandbox tool built by Rob Weidman, Lehigh’s Senior Digital Scholarship Specialist, helped students see into the “black box” of AI and appreciate the value and depth of their writing.

• Dr. Eric Obeysekare

  • We are in the midst of a paradigm shift when it comes to learning how to code with AI.

  • There are things that humans are good at that AI does not replicate. These include critical thinking and problem solving, communication and collaboration, and cultural agility.

  • To be AI-resilient, have students collaborate more. Working in isolation lends itself to turning to AI as the only teammate at hand.

  • It is important to remember that LLMs have a “born on” date. In response, orient students to the future.

  • Know that students have a thirst to learn AI. Work with that! Emphasize guiding over gatekeeping.

  • Teach AI-supported problem-solving in engineering.

• Dr. Suzanne Edwards

  • Her interest in contract grading and contract-based grading was a response to AI an an emergent technology.

  • Her pedagogy emphasizes choice. There are 12 writing prompts a student might complete in a semester. They have the freedom to work on what interests them.

  • One way to be AI-resilient is to emphasize qualitative feedback and conversations between the professor and the student over a numerical score.

  • In her experience, contract grading increases student autonomy, motivation, and responsibility for learning. More feedback and more emphasis on process leads to better student work.

Final Reflections

• There is no one-size-fits-all solution. Context matters: discipline, course level, and student population.