In Fall of 2023, I began noticing something unusual in my students’ essays. It was almost as if they fell into two distinct groups. One set of papers looked highly polished, especially in grammar, but during one-on-one conferences, several students admitted they had relied heavily on AI tools, sometimes even using them to generate full drafts. What surprised me most was that many of these students didn’t believe they had done anything wrong. The other group of papers was the opposite: rough, underdeveloped, and difficult to follow. More than once, I caught myself thinking that these students could have benefitted from running their drafts through an AI tool before meeting with me.
This contrast pushed me to ask a question I suspect many instructors are now wrestling with: if students were taught how to use AI more deliberately and critically, could it strengthen their writing rather than replace it? Instead of banning AI or ignoring it, I decided to experiment with teaching students to use AI as an assistant, not a substitute, during the revision process.
Why I Focused on AI Literacy
Duri Long and Brian Magerko (2020) define AI literacy as “a set of competencies that enables individuals to critically evaluate AI technologies; communicate and collaborate effectively with AI; and use AI as a tool online, at home, and in the workplace” (2). In my own classroom, this definition helped clarify what I was seeing: students were using AI frequently, but not always thoughtfully.
Over time, I realized that students weren’t struggling with access to AI so much as with how to use it intentionally. While the majority of students reported using AI tools, their confidence in how to use them effectively was much lower. This gap suggested that familiarity with AI did not necessarily translate into critical or ethical use.
These patterns convinced me that students didn’t just need rules about AI, they needed guided practice. Rather than banning AI or ignoring it, I designed a revision project that explicitly framed AI as an assistant, not a substitute, and asked students to make intentional decisions about when to accept, revise, or reject its suggestions.
How the Assignment Worked
To address these concerns, I designed an assignment that framed AI as a revision assistant rather than a substitute for student writing. The goal was to help students think critically about AI feedback, reflect on ethical use, and make intentional decisions about which suggestions to accept or reject.
For the assignment, students revisited a previously submitted Character Analysis Research Paper and ran it through the Hemingway Editor, an AI writing assistant that focuses on sentence clarity, structure, grammar, and readability. Students were required to review between three and eight suggested changes and were free to accept them, revise them in their own words, or reject them entirely. The tool also lets students adjust the reading level of their drafts, which led to some of our best conversations about oversimplifying ideas and when clarity starts to flatten a student’s voice. Unlike generative tools, Hemingway presents options rather than producing content.
After revising, students compared their original and revised drafts and wrote a brief reflection discussing what changes they made and how interacting with the AI tool influenced their writing process.
What Students Did with AI
Before beginning the assignment, students showed uneven preparation when it came to AI use. Most could distinguish between generative tools that produce content and assistive tools designed to support revision, while others reported little familiarity with AI writing tools at all. Those with prior experience tended to mention platforms such as Grammarly or ChatGPT, suggesting that while AI was not new to students, their understanding of how—and when—to use it effectively was inconsistent.
Students also brought clear concerns into the project. The most common fears centered on AI generating incorrect information, reducing their individual voice, or crossing ethical lines that felt uncomfortably close to cheating. These concerns echoed what many instructors worry about as well: that AI might shortcut learning or quietly take over the work students should be doing themselves. Rather than dismissing these fears, the assignment allowed space for students to confront them directly through guided use and reflection.
Despite these concerns, many students found that using an assistive tool during revision was helpful. They reported improvements in grammar, clarity, and sentence length, and several noted that the tool helped them see patterns in their writing they had previously overlooked Some students resisted the tool, describing its suggestions as overly simplistic or “robotic.” Even so, many still made meaningful revisions, selecting only suggestions that aligned with their intentions and rejecting the rest. This selective engagement suggested growing critical judgment rather than blind acceptance.
While the tool did not dramatically alter overall performance, it raised an important question for me as an instructor: how might outcomes change if AI were introduced earlier in the writing process rather than after a paper had already been graded? These mixed reactions reinforced that when and how students use AI matters just as much as whether they use it at all.
Moving AI Into the Writing Process
Based on what I learned during the first semester, I revised the assignment for Spring 2025. Although students found the AI assistant helpful, many felt overwhelmed by the site’s volume of feedback, so I continued using Hemingway Editor because it required students to evaluate and select suggestions rather than accept generated content.
Several structural changes shaped the second version of the assignment. Students were given structured, in-class time to work with Hemingway Editor, beginning with a brief demonstration and followed by independent practice. I also adjusted the timing of the assignment so that students used the AI tool before submitting their final drafts, positioning it within the drafting process rather than as a post-assignment reflection. Finally, I added a follow-up survey question asking whether students’ concerns about using AI in academic settings had changed after completing the assignment.
What This Taught Me About Teaching with AI
Comparing the two versions of this project highlighted how much instructional framing shapes student behavior. When AI use was optional and did not affect grades, students experimented with the tool more freely but also more casually. When the AI assistant became part of the graded writing process, students were noticeably more selective, treating the tool as something to consult, question, and sometimes reject rather than blindly follow.
This shift underscored the importance of how AI is positioned in the classroom. Introducing the assistant earlier in the writing process—and making its use consequential—encouraged students to slow down and engage more deliberately with their revision choices, focusing on what aligned with their purpose and voice.
Modeling also mattered. When I demonstrated how to work through the tool transparently, thinking aloud about which suggestions to accept or reject, students became more confident in making their own decisions and less reliant on the tool to “fix” their writing.
In future versions of this assignment, I plan to ask students to experiment with multiple AI tools or use different tools at different stages of the writing process. Giving students opportunities to evaluate AI tools for themselves may further strengthen their critical judgment and ethical awareness.
While this project was limited to a single course, it points to a broader takeaway for instructors navigating AI in their classrooms. Students benefit less from strict rules or blanket warnings than from clear expectations, guided practice, and space to reflect. When AI is treated as a partner in the learning process rather than an enemy, students are more likely to engage actively with their writing and leave better prepared to use these tools responsibly beyond the classroom.
Sara Welshimer is a community college English instructor at Amarillo College who teaches first-year composition. Her work focuses on writing pedagogy, AI literacy, and helping students use emerging technologies ethically and thoughtfully as part of the writing process.
Reference
Long, Duri, and Brian Magerko. “What Is AI Literacy? Competencies and Design Considerations.” Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (CHI ’20), Association for Computing Machinery, 2020, pp. 1–16. https://doi.org/10.1145/3313831.3376727.