When ChatGPT first appeared on the academic scene, I struggled with deciding which path I was going to choose in my own classes, turning to handwritten drafts as a stopgap, but operating with the knowledge that, in the long run, these students may actually need to know how to use AI in their future positions. I discussed detector software, outright bans, and fully embracing AI as possible positions. As I watched my students navigate this new landscape, I realized that prohibition wasn’t the answer. Students are already using AI. The question isn’t whether they’ll encounter it, but how they’ll think about it.
What I needed wasn’t a ban. I needed a path to knowledge and a way for students to make ethical choices.
This three-class lesson series across two weeks emerged from a simple premise: students can’t make ethical decisions about AI if they don’t understand their own thinking processes first. By combining critical readings, hands-on AI experimentation, and deep reflection on reading and writing itself, I’ve created an assignment sequence that transforms students from passive AI users into metacognitively aware writers who can articulate when, why, and how AI supports, or sabotages, their learning.
Here’s how it works.
Class One: The Dark Side First
In-Class Discussion: What’s Wrong With AI?
I begin where students least expect: with AI’s problems. We don’t soft-pedal this. Instead, we dive into:
- Disinformation, propaganda, and the distortion of reality
- The digital divide: How AI is only making it worse
- Plagiarism and privacy issues: What happens to your data when you feed it to a chatbot? The impact on the art fields
- Automated labor: The loss of human jobs
- Climate Impacts: The carbon and water use
- Stereotyping and bias: How does AI perpetuate harmful patterns? Tainted training data, averaging produces cliched results.
- The replacement of thinking: What do we lose when we outsource cognition?
Only after we’ve thoroughly examined the risks do we shift to the possibility: When might AI ethically support learning?
This sequencing is deliberate. Students need to approach AI with critical eyes wide open, not naive optimism.
Homework: Real-World Consequences
Students complete two readings:
- “Your Brain on GPT” – An article grounding our concerns in neuroscience, showing how outsourcing thinking actually rewires our cognitive processes. This isn’t moral panic; it’s brain science.
- A current news article about someone caught using AI inappropriately in academic or professional contexts. Students must find their own source (from reliable news outlets) and answer: “What can you do in your own work to ensure that you use AI ethically and in accordance with academic integrity policies and professional standards?”
By having students research real cases themselves, they become investigators, not passive recipients of warnings. The consequences become concrete, not abstract.
Class Two: Why We Write (Before We Prompt)
Freewriting: Personal Writing Inventory
Before students can evaluate AI’s role in their writing, they need to understand their own. I begin class with a freewriting prompt:
“Why do you write? List all the reasons—for classes, for yourself, for others. How does writing help you? Where does it challenge you?”
This isn’t busywork. It’s groundwork. Students need to articulate their relationship with writing before they can recognize when AI disrupts or supports that relationship.
Homework: “Shitty First Drafts” by Anne Lamott
Students read Lamott’s essay about the messy, non-linear reality of writing. This reading does double duty: it normalizes struggle (writing is supposed to be hard at first), and it sets up our AI experiment perfectly.
Class Three: The AI Experiment
This is where metacognition meets practice.
Opening Discussion: Defining the Line
Before we touch ChatGPT, we discuss:
- What makes AI use “cheating” vs. “learning support”?
- How is asking AI different from asking a study group or tutor?
- When does getting help become avoiding the work?
Students set their own ethical frameworks before experimenting. This primes them to evaluate their experience critically.
Activity Part 1: Comparing Human and AI Analysis
Students prepare by identifying from “Shitty First Drafts”:
- 3 confusing passages
- 2 main ideas
- 1 personal reaction
Then, in pairs or small groups, they submit the Lamott essay to ChatGPT with two prompts:
- “Summarize this article.”
- “What are the main points?”
The comparison moment: Students match their own main ideas against ChatGPT’s. What did the AI catch? What did it miss? Whose interpretation feels more insightful?
Activity Part 2: Learning-Centered Prompts
Next, groups choose one of three more sophisticated prompts:
- “I’m confused about [specific passage]. Can you explain what Lamott means here?”
- “Lamott uses the metaphor of [X]. How does this connect to her larger point about writing?”
- “I think the main argument is [your interpretation]. Does this align with the text, or am I missing something?”
These prompts model curiosity, not abdication. They’re asking for help understanding, not outsourcing comprehension.
Reflection Discussion
After experimenting, we debrief:
- When did ChatGPT help you think more vs. think less?
- Did any AI responses seem wrong or shallow? How did you catch that?
- What’s the difference between “help me understand” and “do this for me”?
Students discover that the quality of the prompt determines the quality of thinking—and that AI can be a mirror, reflecting back the depth (or shallowness) of their inquiry.
Deep Reading Discussion
We transition to discussing Lamott’s essay more deeply:
- What does Lamott mean by “the fantasy of the uninitiated” (paragraph 1)?
- In paragraph 7, she “eventually let [herself] trust the process—sort of, more or less.” Why “more or less”? Is this wariness personal or universal to writers?
- Is writing a first draft more about product or process? Do you agree for your own drafts?
- How can writing serve as self-discovery? Have you experienced this?
These questions are discussed in class and require personal synthesis. That helps to integrate personal stories and prevent the quick AI fix.
The Final Reveal: AI vs. Student Voice
In groups of four, students give ChatGPT this prompt:
“Write a 500-word essay from a college student’s perspective on why you write. Reflect on personal motivations, whether for school, expression, or other reasons. Consider how writing helps you process thoughts, communicate ideas, or explore creativity. Discuss how writing has shaped your identity as a student or impacted your learning. Include specific examples.”
Groups post their AI essays to a discussion board, then compare them to:
- The other groups’ AI-generated essays
- Their own class 2 freewriting about why they write
The critical question: “Which key features suggest this was written by ChatGPT rather than a student?”
Students notice:
- Generic “examples” that could apply to anyone
- Lack of genuine vulnerability or specificity
- Smooth, structurally perfect prose that sounds… hollow
- Absence of authentic voice
The revelation: AI can mimic form, but it cannot replicate lived experience, genuine confusion, or original thought. And most importantly, students can tell the difference when they’re looking for it.
Why This Works
This sequence succeeds because it:
- Builds from critique to curiosity: Students approach AI skeptically but scientifically, not fearfully.
- Centers metacognition: Every activity asks students to observe their own thinking, not just produce content.
- Makes the abstract concrete: Students experience the difference between helpful AI prompts and thought-replacement prompts.
- Honors student intelligence: Rather than lecturing about academic integrity, I trust students to discover ethical boundaries through experimentation and reflection.
- Embraces messiness: By pairing Lamott’s celebration of “shitty first drafts” with AI’s too-perfect outputs, students see that struggle isn’t a bug; it’s the feature that makes learning possible.
What Students Take Away
By the end of three classes, students can:
- Articulate when AI supports vs. replaces their thinking
- Identify AI-generated text by its tell-tale smoothness and lack of genuine voice
- Craft prompts that deepen rather than bypass learning
- Defend their own writing choices with reference to academic integrity
- Recognize that writing is thinking made visible, and that’s worth protecting
Most importantly, they’ve moved from “Is it okay to use AI?” to “Am I thinking, or is AI thinking for me?”
It is metacognition and digital literacy synthesized.
The most revealing moment? When students first realize we’re going to use ChatGPT in class. “Wait—we’re allowed to do this?” “This feels wrong.”
As we work through the activities, something shifts. By the end, students consistently tell me two things:
- “I can see how it can help with studying.” They recognize AI’s value as a thought partner—when used to deepen understanding, not replace it.
- “But I don’t want to use it to complete assignments.” Not because they fear punishment, but because they’ve experienced the hollowness of AI-generated work compared to their own authentic thinking.
This isn’t compliance, it’s wisdom. Students learn to ask themselves: Will this help me grow, or help me avoid growing? That question will serve them far beyond my classroom.
Resources Needed
- “Your Brain on GPT” (or similar neuroscience article on AI and cognition). I like this one because the authors made it difficult for AI to summarize the article.
- “Shitty First Drafts” by Anne Lamott
- Access to ChatGPT or other AI (free tier works fine)
- Discussion board for posting AI essays
Final Thoughts
Our students will use AI. That’s not a crisis; it’s a reality. Our job isn’t to prevent that use, but to ensure they approach it with clear eyes, ethical frameworks, and self-awareness.
This lesson series teaches students to think about their thinking, to value their own cognitive labor, and to recognize that the messiness of learning, the confusion, the struggle, the shitty first drafts, is where growth happens.
And that’s something no chatbot can generate for them.
*AI Disclosure: This blog post was conceived and written by a human, but refined and organized using Claude AI.