TLC partners with the University Writing Council (UWC) to share pedagogical strategies for teaching writing. Today’s author, Jaeyong Choi is a member of the UWC Advisory Board. This is the third in a series of posts addressing writing-intensive classes in the age of generative artificial intelligence. The previous post discussed “Three Approaches to AI Integration with Writing Courses.”

When generative AI first became widely available to the public, my initial response as a professor was to prohibit its use. I wanted time to observe how the technology would affect student learning before revising my course policy. By 2023, however, it became clear that a restrictive approach was not working. Regardless of my policies, students were using AI for writing assignments. Even after I introduced additional safeguards, such as requiring timestamped references after documentaries (minute and second) or requiring specific page citations for readings, I could not realistically eliminate AI use.

This blog post explains how I am approaching that shift. Before I came up with the idea of structured, stage-specific guidance for AI- supported writing in class, I reviewed Scholarship of Teaching and Learning (SoTL) research to understand how other faculty are using AI as a teaching tool and what outcomes have been documented. Existing studies suggest that even brief, instructor-led guidance can improve immediate outcomes such as task performance, creativity, engagement, or ethical awareness when students use AI for a single assignment or short intervention (Abuadas et al., 2025; Bushnell & Harrison, 2025; Kumar & Sangwan, 2024; Yan & Zhang, 2024). However, most studies treat AI as an isolated tool rather than as part of a recurring, scaffolded workflow embedded across multiple stages of writing over several weeks. Prior work also tends to focus on short-term outcomes such as essay quality or satisfaction rather than longer-term development of writing efficacy and AI literacy.

In response, I developed a semester-long, multi-stage protocol that integrates AI into an authentic writing course in a structured way. My goal is to increase students’ writing efficacy and AI literacy while maintaining academic integrity and strong disciplinary learning. Through the Pedagogical Innovations Faculty Learning Community supported by the TLC and the ORSP, I am currently evaluating whether this structured, stage-specific approach is associated with changes in students’ writing efficacy and AI literacy.

Course structure and scaffolded writing tasks

The protocol begins with scaffolded writing assignments. For example, in CRJ316 Terrorism (a writing emphasis course), students complete a final research paper that applies a criminological theory to a specific terrorist organization. Students are randomly assigned one criminological theory and one terrorist organization. They also receive one peer-reviewed article connected to their assigned theory and two peer-reviewed articles connected to their assigned organization, and students are assigned different articles. In addition, they must locate two additional peer-reviewed articles on their own. Before drafting the final paper, students read five articles and complete an annotated bibliography. This sequencing is designed to build content knowledge, improve reading comprehension of research articles, and support a stronger final paper.

Stage 1: AI literacy session before the annotated bibliography

Before students begin reading and writing the annotated bibliography, I hold the first AI literacy session: “What Is GPT? Understanding How AI Chatbots Generate Responses.” In this class, students learn what the terms generative, pre-trained, and transformer mean in GPT models. They also learn how chatbots generate text through next-word prediction, how pretraining supports language pattern learning, and how context shapes meaning in AI-generated responses.

A key activity in the first AI literacy session involves comparing a human-written annotated bibliography entry with an AI-generated version. Students evaluate both using a rubric designed to highlight meaningful differences, such as accuracy of interpretation, use of specific article details, and analytical depth versus summary.

Structured, Stage-Specific Workflow for AI-Supported Writing Assignments and Embedded AI Literacy Instruction

The purpose is to help students recognize that generative AI can produce convincing text that is vague, inaccurate, or not grounded in the article. This activity is intended to motivate students to read the original research and take ownership of their writing.

Stage 2: AI literacy session during outlining

After the annotated bibliography, students develop an outline for their final research paper based on what they learned from the articles. At this stage, I deliver the second AI literacy session: “Prompting as a Skill: Communicating Clearly with AI Systems.” In this session, students practice using a structured prompt framework (Task, Format, Voice, Context) and learn how to refine outputs through iterative feedback. Students then use AI tools to request outline feedback and revision suggestions, with the expectation that they evaluate and revise those suggestions rather than copy them.

Stage 3: AI literacy session on ethical use during drafting

Shortly after students begin drafting their final paper, I provide a third AI literacy session: “Ethical AI Use in Academic Writing.” Students examine ethical and social issues related to AI, including fairness, accountability, and transparency. We also discuss environmental impacts and strategies for responsible use. In addition, students learn how to acknowledge AI use appropriately in academic work.

Stage 4: AI Use Reflection

The final component is an AI use reflection designed to strengthen writing development through metacognition. Students reflect on how AI supported their writing, where it fell short, and what they learned about responsible technology use. Students also identify all stages where they used AI tools, including brainstorming, outlining, drafting, revising organization and clarity, and editing grammar or APA formatting. They explain what they asked AI to do, what they hoped to achieve, and which stage was most helpful and why.

I received IRB approval to evaluate whether this structured, stage-specific guidance changes students’ writing efficacy and AI literacy using pre- and post-course data (IRB-FY2026-103: The Impact of Structured, Stage-Specific Guidance for AI-Supported Writing on Students’ Writing Efficacy and AI Literacy). I am looking forward to examining the results.

Clearly, my approach using structured, stage-specific guidance for AI-supported writing is only one of many possible ways to improve students’ AI literacy and support their effective and ethical use of AI. I look forward to seeing additional evidence-based teaching strategies shared through the TLC blog.

Author Note: Portions of this manuscript were edited with the assistance of ChatGPT (OpenAI, 2026) to improve clarity and conciseness. The author reviewed and revised all edited content for accuracy and coherence.

References

Abuadas, M., Albikawi, Z., & Rayani, A. (2025). The impact of an AI-focused ethics education program on nursing students’ ethical awareness, moral sensitivity, attitudes, and generative AI adoption intention: A quasi-experimental study. BMC Nursing, 24, 720. https://doi.org/10.1186/s12912-025-03458-2

Bushnell, J. T., & Harrison, W. (2025, April). A new muse: How guided AI use impacts creativity in online creative writing courses [White paper]. Oregon State University Ecampus Research Unit. https://ecampus.oregonstate.edu/research/wp-content/uploads/Bushnell-Harrison-2025.White-paper.pdf

Kumar, A., & Sangwan, S. R. (2024). Conceptualizing AI literacy: Educational and policy initiatives for a future-ready society. International Journal of All Research Education & Scientific Methods, 12(4), 1543–1551.

Yan, D., & Zhang, S. (2024). L2 writer engagement with automated written corrective feedback provided by ChatGPT: A mixed-method multiple case study. Humanities and Social Sciences Communications, 11(1), 1–14.