Strategies for Designing AI-Resistant Assignments

It may not be possible to create assignments that are completely immune to AI use, but there are strategies that you can use to discourage it and increase student motivation to do the work without AI assistance.

 

Encourage Effective Learning Strategies

Offer students guidance on effective learning strategies, such as retrieval practice, interleaving, and metacognition, as they may be unaware of these evidence-based learning techniques. For example, students often report feeling like they learn more from passive lectures than from participating actively in class, even though their performance is worse in the former condition (Deslauriers, et al., 2019). 

Remind students that relying on AI tools may hinder their learning and development of fundamental skills that they’ll need to use in the future. Students may be tempted to use shortcuts, including generative AI, to maximize efficiency or decrease time spent on assignments. Unsurprisingly, students who reported copying their homework did not perform as well on exams as those who reported doing their own work (Glass and Kang 2022).   

Russell Johnson, from the Divinity School, argues why doing your own writing is so important in his article, “On ChatGPT: A Letter to My Students”:

“The reason why I insist you do writing assignments like this is that they give you valuable practice discovering insights and communicating them to others. Reading texts closely, encountering a problem, developing a plausible interpretation, and persuading readers of that interpretation—these are the steps one must go through in order to write a good paper. Going through these steps again and again makes us clearer thinkers and better communicators.”

For more on Johnson’s approach to AI in his teaching, see his Teaching Spotlight from the Chicago Center for Teaching and Learning.

 

Slow Down Reading

Easily producing reading summaries and “chatting with a text” using AI may help students initially encounter a difficult text, but it also may offer an illusory mastery of the material if used uncritically, as Marc Watkins argues in Rhetorica, his Substack. Encourage students to slow down their reading of critical texts in your discipline by engaging with them meaningfully.

Social Annotation

Have students engage with texts, one another, and yourself using a social annotation tool like Hypothesis. As Waktins notes, “Creating active reading assignments where students are required to annotate a text and discuss it with fellow students slows down the process and adds friction to the experience. It also makes the reading a social activity as opposed to an individual one and invites inquiry and debate within the reading process.” You may point out to students that they can offer glosses, provide interpretations, ask or respond to questions, and offer personal responses in their annotations, as appropriate.

Commonplacing

Have students record crucial quotes and interact with important passages in their own commonplace books or notes, which the UChicago Library characterizes as “at once a book form and a method of reading.” Popular among writers and thinkers, from John Locke to the present day, commonplace books are typically organized by theme and include quotes, ideas, and observations that are often useful for future reference and reflection. This can be done on pen and paper, in Google Docs, or in more purpose-built tools like Microsoft OneNote and Notion.  

Concept Maps

Concept maps can be powerful tools for students to engage critically with texts, lectures, and other class activities. Students can create diagrams that show the relationships between ideas, with connecting lines and linking words visually and textually describing the relationships between concepts. Learn more about concept mapping and how it’s being used by UChicago instructors, including Jennifer Spruil, in the Social Sciences Core of the College.

Focus on Process

Focusing on process—e.g., the steps you take to solve a problem or write an essay—may become more important as generative AI can quickly produce fluent, albeit generic and possibly inaccurate, text. Demystifying how academic “products” are made—e.g., identifying the recursive processes involved in writing an essay, or common steps in deriving an equation—and engaging students in reflection on their own processes, may help students avoid the temptation of solely and uncritically relying on AI tools.  

Scaffold Writing Assignments

Break up large writing assignments into smaller constituent parts. For instance, for a research paper due at the end of the quarter:

  1. Incorporate time for in-class, exploratory thinking and writing; you may wish to collect samples of students’ writing to get a sense for how they write. 

  2. Assign students to submit a topic and draft thesis for your feedback. 

  3. Have students develop and submit a literature review and outline. 

  4. Meet with students briefly to discuss their progress.

  5. Ask students to provide feedback to one another before submitting final drafts, if possible. 

  6. Allow students the opportunity to revise and resubmit following your final round of feedback.

You may also ask students to complete these tasks using Google Docs or Microsoft 365, so you can refer to the version history if desired. 

To learn how an instructor in the Social Sciences Core in the College implements this kind of approach, read Sarah Johnson’s “Scaffolding the Writing Process: An Approach to Assignment Design in the SOSC Core.”

Require Student Reflection

Require students to reflect on their own thinking, processes, and approaches in their work. This kind of reflection can help promote metacognition, which can help students become better regulated learners who are able to identify gaps in knowledge and opportunities for continued learning. For example, you might ask students to:

  • Explain why they chose the references and other source materials for their projects.

  • Describe their approach to solving a problem or completing a project, including identifying barriers or false starts along the way.

  • Identify any lessons learned that can be applied to future contexts.   

  • Discuss how a concept relates to their own experience, if applicable, and how understanding the concept may influence their interpretation of the experience.

Encourage Active, Experiential, and Collaborative Learning

Active learning engages students in the learning process, rather than passively listening to a lecture or reading a text. Use software like PollEverywhere to prompt students to respond to questions in real time. You may wish to intersperse your lecture with knowledge checks and opportunities for retrieval practice, or assign students “exit tickets” prior to leaving class.

Experiential learning engages students through reflection on experiences, often outside the classroom, such as in field studies, lab work, and supervised internships. GAI models do not have direct access to students’ experiences and are unable to make meaning from them.

Collaborative learning involves students working together to achieve a common goal. One subtype of collaborative learning, called project-based learning, where students collaborate with one another on real-world projects, has been shown to positively impact student achievement (Chen and Yang  2019). Students engaged in collaborative projects may be less likely to engage in the unethical use of AI, as they feel accountability not just to their instructors but also to their peers.