Introduction

With the advent and development of ChatGPT and other Generative Artificial Intelligence (GenAI) tools, faculty are faced with new teaching challenges, including determining how these technologies work and how students may or may not use them in class. During the past few years, there has been much advice from educational technology companies and in the popular press about how to deal with GenAI in higher education. This advice often lacks evidence of effectiveness and has sometimes harmed students (for example, blue-book only approaches or relying on GenAI detectors).

Instead of establishing a one-size-fits-all approach, this blog assumes that faculty are in the best position to decide which evidence-based pedagogical strategies will help students meet course learning outcomes. These strategies are not designed to “AI-proof” assignments; rather, the term “AI-resilient” is used to indicate assignments for which students are less likely to use GenAI in non-sanctioned ways. An AI-resilient assignment “ensures that core learning outcomes cannot be easily outsourced to AI—not by relying on students to comply— but by thoughtfully creating conditions and structures that make it hard for students to use AI to complete the core learning tasks” (Digital Education Council).

Policy Assumptions

The MLA-CCCC Joint Task Force on Writing and AI recommends a tiered approach to Generative AI in which college leadership, with input from the faculty, establishes “broad guidelines for GenAI use that the department must follow as they create policies for classrooms. At the same time, individual instructors maintain autonomy by setting AI policies in their courses according to disciplinary knowledge and assignment types but in alignment with guidelines from the department and the administration.” In keeping with this tiered approach, faculty should develop policies on their syllabi and assignments that describe how students may or may not use GenAI. These policies should be consistent with the university’s academic honor code and with guidance from IT on acceptable and ethical AI use. According to the MLA Join Task Force, “Policies should support the development of critical AI literacy rather than resort to blanket restrictions on access.” Moreover, faculty should serve as role models for students by using GenAI in ethical ways and by disclosing any use of AI in their own teaching materials.

Evidence-based teaching strategies

• Transparent writing assignments
  • Specify the task, the purpose of the writing assignment, an authentic audience (preferably, not solely the instructor as the expert), and the criteria for success. Evidence shows that transparent writing assignments help all students succeed. Moreover, students are less likely to use GenAI in inappropriate ways if they know why they are writing, what they are supposed to do, and how they will be evaluated.
• Meaningful assignments
  • Integrate on or more of the following elements: student choice, new kinds of writing assignments, working with others, time on task, personal connection, researching to learn, and integration of the writing process. Students are much less likely to take short cuts on assignments that are meaningful to them.
• Low-stakes writing (in or out of class)
  • These assignments encourage students to demonstrate what they know or to identify areas for future learning in a low-pressure environment. Because these are ungraded (or graded for a small amount of credit), students are less likely to use GenAI in non-sanctioned ways.
• Peer review
  • Peer review allows students to give and receive feedback so that they learn more about their own writing and how their peers deploy writing strategies. Peers can help students identify misuse of GenAI in a less threatening way. Moreover, students are less likely to take shortcuts when they know that peers will be reading their work.
• Conferencing with students
  • Individual conferences allow instructors to support students as they progress through the writing process. Students are less likely to use GenAI as a shortcut if they know that they will be discussing their writing and reading process. In larger classes where conferencing is not possible, peer review or the writing center can afford a similar dialog about writing.
• Reflective writing assignments and writing portfolios
  • Asking students to reflect on what they have written is an effective way to prompt them to think about their writing choices and their growth as writers. Students can also reflect on their own writing compared with writing generated by an AI tool.
• Teaching disciplinary reading strategies
  • Students often struggle with writing assignments because they haven’t understood the assigned reading. Providing students with strategies to read in your discipline (for example, how to close read a literary text or skim a science article) can help students avoid turning to AI for help.

Teaching Strategies with Mixed Evidence of Effectiveness for Student Learning

• In-class written exams
  • While in-class written exams have a long history in the classroom, there is little evidence that these help students learn. Moreover, if these exams are meant to be handwritten (in blue books or other media), know that some of your students will be unable to complete the task without assistive technology.
• Process Tracking
  • This approach involves tracking how students have created and developed a document. Instructors use readily available tools (e.g., Google Docs or Word) or subscription services. Some writing studies scholars worry that process tracking causes students to feel surveilled and “could undermine trust and egalitarianism” (Losh). Others have argued that it is “a much better approach than AI text detection, which is inaccurate, likely biased, easily circumvented, and not transparent” (Mills). These scholars assert that process tracking can work for students if implemented thoughtfully. Be careful that you are not significantly increasing your workload if you decide to adopt process tracking.

Teaching Strategies that Harm Students

• Using GenAI detectors
  • Despite claims by educational technology companies, these so-called detectors produce false positives, and they tend to flag multilingual writers more frequently. Students who are savvy about GenAI can trick AI detectors. Detectors also have the potential to expose student data and intellectual property.
• Assuming that some students won’t be interested in the class because of GenAI
  • Research on teaching and learning has shown that any student can learn and become interested in any subject—if instructors thoughtfully engage students and show them that they can succeed in the course, regardless of individual backgrounds. Courses are more likely to engage students if they prioritize accessibility, engage students in a variety of ways, present information in multiple formats, and allow students some choice in demonstrating what they know.
• Overuse (or misuse) of AI by the instructor
  • Students have reported dissatisfaction with courses that overuse AI in creating assignments and in communicating with students, especially when the use of AI is not disclosed by the instructor. In most cases, AI should not be used to grade student work or to write letters of recommendation.

Sample AI-Resilient Assignments

Assignments with some AI use encouraged

• Create a podcast
• Design a multimedia presentation
• Evaluate/analyze/compare AI outputs
• Design and iterate prompts for an AI tool
• Discuss or analyze an issue related to AI

Assignments with AI use discouraged

• Create and deliver an oral presentation
• Annotate a course reading with classmates
• Create a portfolio and reflect on development of writing skills
• Engage in low-stakes, writing-to-learn exercises

Resources

“Actionable AI in the Humanities Classroom” (MLA-CCCC Joint Task Force)

“An Introduction to AI Policies” (MLA-CCCC Joint Task Force)

Bean and Melzer, Engaging Ideas: The Professor’s Guide to Integrating Writing, Critical Thinking, and Active Learning in the Classroom

Bowen and Watson, Teaching with AI: A Practical Guide to a New Era of Human Learning

Digital Education Council, “The Next Era of Assessment,” 2025

“Student Guide to AI Literacy” (MLA-CCCC Joint Task Force)

“What is Process Tracking and How is it Used to Deter AI Use?” (MLA-CCCC Joint Task Force)