AI in the Classroom is a series highlighting the experiences of Ohio State faculty who are using AI in their teaching — what’s working, what’s still unfolding and where it’s headed next.
Susan Lang is a professor in the Department of English and director of the Center for the Study and Teaching of Writing (CSTW) in the College of Arts and Sciences. As CSTW director, Lang leads university-wide and interdisciplinary initiatives that integrate pedagogy, technology and institutional infrastructure to support writing across the university. Her research examines how digital technologies — including AI-enabled feedback systems and writing analytics — can extend human-centered writing pedagogy in undergraduate and graduate education, particularly in STEM and professional fields.
How should faculty be thinking about AI in courses that involve student writing?
On many campuses, the first response to AI in writing-intensive courses has been to treat it as an academic integrity problem: How can faculty tell whether students are using these tools to complete assignments? That concern is understandable, but it risks starting in the wrong place. The more important question is what students must actually do to complete an assignment.
For decades, many writing assignments worked because producing text required a level of sustained intellectual effort. Students had to interpret a question, analyze sources, weigh evidence, make claims, and decide how to structure an argument. Generative AI weakens that connection. Students can now produce fluent, plausible prose without engaging in many of the cognitive processes writing assignments were designed to promote. Consequently, a finished report, proposal, or essay is no longer reliable evidence that a student has engaged in analysis, judgment, or decision-making.
The challenge becomes designing assignments where those forms of engagement remain necessary. In practice, that might mean asking students to justify interpretations, explain how they evaluated evidence, or produce artifacts that make their decision-making visible.
AI hasn’t created these challenges; it has exposed assumptions built into many of our assignments. We often operated on the premise that if students produced writing, they must have done the thinking required to produce it. That premise was always imperfect. Generative AI simply makes the gap harder to ignore.
In that way, AI functions less as disruption than as diagnostic tool.
How can AI be useful in writing-based assignments, and what should faculty be thinking carefully about when using it?
AI can be useful when it supports thinking rather than replaces it, especially when students are still learning how a discipline constructs knowledge, evaluates evidence, and develops expertise.
Students have long relied on tools that reduce the burden of particular tasks. A calculator allows a student to focus on solving a problem rather than performing arithmetic by hand. AI can serve a similar role in writing by helping students generate possibilities, test ideas, or receive feedback.
When designing assignments, faculty must decide which parts of a task can be supported by AI and which remain the student's responsibility. AI might help generate possible claims, but selecting and justifying the strongest one should remain the student's responsibility.
The goal is to ensure that AI expands students’ options without assuming responsibility for their decisions. AI can expand the range of possibilities available to students, but the educational value emerges when they must evaluate those possibilities, justify their choices, and explain why one path forward is stronger than another.
One concern people raise is that AI may reduce students’ critical thinking or creativity, or make it harder to know what students are learning and producing independently. How should faculty think about that concern when designing writing assignments?
This concern is real, but I would reframe it. The issue is less that AI reduces critical thinking than that it exposes a longstanding design problem in writing instruction. Many writing assignments assume that if students produce writing, they have engaged in the thinking required to produce it. The question is whether prose still provides meaningful evidence of analysis, judgment, and decision-making.
Consider an assignment that asks students to analyze a reading or explain the significance of a concept. Historically, producing a response required students to select evidence, organize ideas, and articulate a position. Today, students can generate plausible responses in seconds. The key design question is what happens next. Does the assignment require the student to interrogate that response, evaluate its accuracy, defend its reasoning, and revise it in light of competing interpretations? Or does it stop at the production of a coherent answer?
If so, the assignment may be measuring text production instead of engagement with ideas. Otherwise, AI can make the intellectual work more visible by shifting attention from text production to evaluation, justification, and revision.
Seen this way, concerns about critical thinking and creativity point less to a technology problem than to a design problem. They call for assignments that make students' reasoning, decisions, and revision processes visible rather than assumed.
As AI becomes more common, what parts of the writing process remain especially important for students to do themselves?
As AI becomes more common, students still need to understand writing as a series of decisions.They must determine whether evidence supports a claim, whether an explanation is accurate, whether a paragraph advances an argument, and whether a piece of writing achieves its purpose. These are not mechanical steps; they are the decisions through which writing becomes a mechanism for learning.
AI can generate possibilities, suggest explanations, and flag areas for revision. It cannot decide what is true, persuasive, or appropriate in context. Those decisions require disciplinary knowledge, rhetorical awareness, and the ability to evaluate competing options.
The goal is to ensure that students remain responsible for selecting, evaluating, and refining ideas. If students use AI to expand their options while retaining responsibility for those decisions, writing assignments can still produce deep learning. When those decisions are outsourced, the core educational value of the activity begins to erode.
What’s one question faculty should ask themselves when deciding how AI fits into writing assignments?
Before making decisions about AI, ask: What kinds of thinking does this assignment require, and how will I know students engaged in that thinking?
Then ask: Which parts of this task can be supported by AI, and which parts must remain the responsibility of the student?
Those questions shift attention from the tool to the learning the assignment is intended to produce.