The availability of generative AI requires writing instructors to be more deliberate about assignment design. Yet many principles we have always valued remain the same: Prompts should provide opportunities for students to use writing as a means to practice critical thinking and reflection; to engage deeply with texts, using sources to help them generate interesting questions and develop compelling arguments; and to recognize the ethical and social dimensions of writing.
We can encourage students to embrace these opportunities through tapping their intrinsic motivation—striving, as John Bradley puts it, to craft assignments that are “transformational” rather than “transactional.” We begin with some general recommendations that draw on Mary-Ann Winkelmes’s research-based guidelines for transparent assignment design and are oriented toward this goal.
Writing assignments should include the following, ideally in about one page:
Purpose: Begin by briefly describing the purpose of the assignment, how doing the assignment helps students reach a particular learning goal of the course. You may want to connect this goal to a future application in or beyond your course.
Tasks: Craft a concise, specific prompt that is meaningful and relevant to students’ lives. This might mean asking students to draw on personal experiences, but it might also mean asking them to write for an authentic audience (not just you), in a genre relevant to their future professional or civic lives, or in connection to an urgent local or global challenge. Emphasize “distinctive human skills” as described on this useful chart, Bloom’s Taxonomy Revisited.
Be sure to incorporate a series of process steps and deadlines. This might include brainstorming activities, proposals, drafts (for peer review, in-class workshops, or conferences), and reflective writing. Emphasize oral and interactive activities.
Criteria: Show students what success looks like. Offer models of effective compositions, and invite students to analyze those models to develop a critical understanding of how they work. Consider using models published in Deerfield, the Writing Program journal of outstanding undergraduate writing, or other BU student publications.
Tell students what you will focus on in your feedback and assessment. Consider deemphasizing criteria that AI-generated prose can easily meet and emphasizing criteria like intellectual risk-taking, originality, nuance, etc. (Note the useful distinction Kevin Gannon makes between “logistical rigor” and “cognitive rigor”—and don’t mistake the former for the latter.)
See also Anatomy of an Assignment Sheet.
Crafting Assignments That Discourage AI Use
The suggestions above should help you craft writing assignments that motivate students to learn, but there are some additional measures you can take to encourage students to engage in the writing process and discourage the use of AI shortcuts.
- Make sure that “distinctive human skills” are front and center in each step of the assignment.
- Ask students to write in longhand and/or talk about specific quotes or sources in class as part of the writing process.
- Have students record their reading and research process. Consider assigning a research log or using a social annotation tool like Perusall. Formal or informal stepping-stone (scaffolded) assignments can help to ensure that students are engaging in authentic research and reading deeply.
- Offer opportunities for students to write about their own lives and experiences, as appropriate to the assignment genre and course topics.
- Value creativity and difference. Invite students to explore nonstandard language and question genre conventions in a way that is relevant to your topic.
Crafting Assignments that Incorporate Generative AI
If you are interested in preparing students to use these new writing technologies well, here are a few suggestions for incorporating generative AI into writing assignments in a way that supports learning.
- Be explicit about the kind of AI use you are authorizing for the assignment, and model effective prompting for this use. Chapter 3 of Bowen and Watson’s Teaching with AI offers a useful overview.
- Consider an assignment that showcases what AI can do well and where it falls short–for example, a two-part assignment that asks students to use AI to draft, then write a critical analysis of what AI does well and poorly before they go on to revise or discard the draft.
- Engage AI in the revision process after students have drafted an essay. Let students experiment with the ways that AI can help them reformulate, rephrase, or reorganize their ideas. Include opportunities for reflection regarding their experience working with AI as a collaborator.
- Remember that servers that run AI, such as ChatGPT, are not always accessible during times of high demand, and plan ahead if you intend to use AI live in the classroom. For instance, if you are planning a live demonstration of AI, have it generate one or two responses to be used as backup in the event that the AI tool is not accessible during your class meeting.
- Be explicit about how you want students to cite AI. For example, refer to the MLA guide on Citing Generative AI in MLA Style.
More ideas that can be incorporated into the scaffolding for a writing assignment are available here.
Responding to Unauthorized Uses of AI for Writing
“Use caution about responses that emphasize surveillance or restrictions on the writing process that make the conditions of writing for class radically different from writing conditions students will encounter in other classes, work environments, and their personal lives.”
—MLA-CCCC Joint Task Force on Writing and AI
Unauthorized use of generative AI tools for academic work may fall within the Academic Conduct Code’s definition of cheating: “Any attempt by students to alter their performance on an examination or assignment in violation of the stated or commonly understood ground rules.” However, documenting evidence of this kind of misconduct is not easy. The results of detectors like GPTZero and Turnitin are not very reliable, with evidence of “an alarming bias against non-native English speakers.” (Concern about this bias is the basis for the Writing Program’s authorization of AI use for grammar, usage, and vocabulary in all WR courses; refer to our current guidelines in Section 2 of the syllabus templates.)
Therefore, if you are concerned that a student has used AI to do their work in a way that violates academic integrity, it’s best not to accuse a student based on detector results. Instead, invite the student to have a conversation with you about their process and their ideas. Be honest about your concerns and allow the student an opportunity to respond.
Learn More: Writing Instruction in the Age of Generative AI
