Morgan Harms-Abasolo
University of Nebraska Omaha
In this lesson, first-year writing students learn to engineer ChatGPT prompts to create an artificial intelligence (AI) Audience Avatar that functions as a mock focus group for their upcoming multimodal argument. As students interview their simulated audience members, they identify biased or stereotypical language to foster critical awareness of how AI simulates audience identity. The activity supports rhetorical knowledge transfer by helping students adapt their arguments to new audiences, genres, and purposes. This lesson can be adapted for any class level focused on rhetoric, multimodal composition, and communication.
Learning Goals
Original Assignment Context: At the end of a sixteen-week first-year writing course, students are remixing their standard academic papers into a multimodal genre of their choosing for a new target audience.
Materials Needed: Large language model, such as ChatGPT-4o
Time Frame: 1 hour 15 minutes
Overview: In my first-year writing course, students undertake a major two-part assignment. In the first part, students write a standard academic argument paper appealing to stakeholders to take action on a social justice issue impacting their local community. In part two, students remix their academic paper into a multimodal project targeted at their community members.
Transitioning between these genres and audiences often poses challenges for students. To facilitate this shift, I developed this 75-minute lesson introducing the concept of an "AI Audience Avatar," generated using ChatGPT. This activity uses AI to help students gain a deeper understanding of their new audience by simulating a focus group-style dialogue. It encourages them to interrogate an AI-generated avatar and reflect on adapting their argument's genre, purpose, and rhetorical strategies for their new audience. It also provides an opportunity for students to critically evaluate the AI’s output for problematic content. This encourages students to read with a more critical eye, which, in turn, strengthens their ability to identify hallucinated content. This lesson supports both rhetorical awareness and adaptability, as well as ethical awareness, in digital writing contexts.
I have taught this lesson for three semesters. Over these semesters, I have realized that it is essential to repeatedly walk students through the “Students’ ‘Perfect’ Prompt Formula” while color coding an example, as writing AI prompts is similar to learning a new writing genre. For their final multimodal project, students are only graded on their cover letters. In these cover letters, students who reflect on regularly returning to their avatar often outperform those who do not and report increased confidence in their use of rhetoric.
1. Introduction to AI Audience Avatars (10 minutes)
I begin the lesson by defining an AI Audience Avatar as a personalized, simulated representation of an audience member. I explain that the avatar serves as a focus group, allowing students to personally get to know their audience and inform their rhetorical choices. I describe that, just as a business would run a focus group for a new product, we, too, must get to know our audience before we can plan how to persuade them.
To illustrate the risks of poor audience awareness, we discuss how forty-four-year-old Kim Kardashian misjudged her audience by targeting college-aged women instead of her actual audience, peers her own age. Her audience disconnect and immature actions alienated both groups, offering students a relatable example of how misreading an audience can result in rhetorical failure.
At this stage in the semester, students have already learned about larger AI ethical concerns such as privacy and originality, which I recommend other instructors do before attempting the AI Audience Avatar. In this class period, I emphasize that ChatGPT can reproduce biased, offensive, or outdated output due to the model’s training data. I encourage students to read with a critical eye and always to be skeptical of output from AI. This discussion aligns with critical digital pedagogy, which encourages educators and students to critically assess the tools they use and the content they produce. As Maha Bali notes, cultivating critical AI literacies involves understanding and questioning the biases embedded in AI technologies (Bali, 2024).
2. Teaching Prompt Engineering and Live Demo (25 minutes)
Next, I introduce the “Students’ ‘Perfect’ Prompt Formula”: [Task] + [Format] + [Tone] + [Context], always ending with, “Before you generate your response, what questions do you have?” This formula originates from YouTube AI content creator Jeff Su. I have found YouTube to be the most helpful source for learning AI skills due to its ability to share screen recordings of AI tools in use. While Su’s prompt formula has yielded excellent results in my personal use, I have simplified it from the suggested six components to four components for the ease of novice users, especially considering that first-year students often become frustrated writing lengthy AI prompts. The [Task] is the goal of the assignment; [Format] is the genre of the output; [Tone] is the writing voice and style; and [Context] is the assignment details, examples, or grading criteria. Prompting the AI to ask the user clarifying questions at the end of a prompt is imperative, as it adds extra context and creates more accurate output.
For this assignment, I prompt the AI to produce its results in both diary entry form and Q&A style, allowing students to see a sample of the simulated audience’s possible perspective and writing style while also receiving direct answers. As I model this for students, I color-code the formula and my example prompt, which helps increase students’ comprehension of how to apply it.
As a class, we engineer a ChatGPT prompt requiring it to write in Gen Z slang. During the live demo, ChatGPT typically generates outdated terms, such as “YOLO,” a phrase popularized by Millennials in 2011. While “you only live once” is not offensive, this linguistic misrepresentation highlights AI’s limitations and serves as an ethical checkpoint for students, reinforcing the need for rhetorical fine-tuning.
3. Students Engineer Prompts and Annotate AI Output (20 minutes)
Using the “Students’ ‘Perfect’ Prompt Formula,” students then draft their own ChatGPT prompts tailored to their multimodal project topics and copy and paste set questions for the avatar. As I introduce the questions for the avatar, we have a class discussion about how the avatar’s response to these questions can inform their argument. The students’ prompts include the following questions:
Students must be logged into their ChatGPT account before entering this prompt, as the AI model will save this conversation, allowing students to regularly return to their avatar to ask it for feedback on their rhetorical choices as they work on their project. Once students have entered their prompt into ChatGPT, they annotate its output while asking themselves, “How can I adapt my argument, its context, and its purpose to better meet my avatar’s needs?” Of course, students also need to identify any unethical output as they read and annotate.
4. Reflect and Adapt (20 minutes)
Finally, without utilizing AI, students write their responses to five reflection questions that are designed to promote rhetorical transfer from their academic argument paper into their multimodal project:
For this lesson, students are only graded on the Reflect and Adapt portion of the assignment. Here, I am looking for evidence of the student’s critical thinking and that they are planning adaptations for their new audience and genre. This reflection establishes a crucial bridge between the academic argument and the upcoming multimodal remix, facilitating students’ transfer of rhetorical concepts.
My thanks to Rebekah Hayes for her thoughtful revision suggestions. This lesson draws inspiration from Angela E. Lauria's The Equalizing Quill: 6 Ways Generative AI Can Boost Publication Access for Underrepresented Authors (2023), which emphasizes AI's potential to democratize writing. It was also informed by Maha Bali’s appearance on the Teaching in Higher Ed podcast. I owe my thanks to Jeff Su’s YouTube video “Master the perfect ChatGPT prompt formula (in just 8 minutes)” for teaching me how to prompt the AI effectively