Dear Student,

I’m sipping coffee as I write this. I make it in a 16-ounce thermos each morning, grinding locally roasted beans, stuffing the grounds into a stainless steel pod for the Keurig (my coffee machine), then using a frother to mix in raw cream and 2½ teaspoons of turbinado sugar. Mmmmm.

If I’m away from home, the options change. I’ll look for the most hipster coffee shop nearby and splurge on their largest almond milk cappuccino. Often they’ll ask if I want to add flavoring, and I’m tempted to gasp from shock:

“And ruin Mother Nature’s most extraordinary flavor with a lab-produced, artificial trick?”

But I tamp down the impulse to berate the barista. Instead, I shake my head no and tell them I love the OG, thanks anyway.

Well, as a teacher and a writer, I have multiple objectives behind every choice I make, whether in the classroom or on the page, so let me elucidate.

1. Making coffee is a process―and a personal one―that requires experimentation to decode your preferences, much like with the writing process. Throughout either process, various people and/or tools might help you achieve the final product, and I realized in a sudden spark of inspiration that coffee is a useful analogy for understanding the help we might get from chatbots (i.e. generative artificial intelligence such as ChatGPT).
2. I described my own coffee preferences partly because I believe you’re more likely to hear me out if I exhibit some personality. (To those in my class who already know my quirky self, Hi friends!!! *waves enthusiastically*)
3. I want you to know immediately that a human is writing this.

A chatbot has no coffee preferences. It has never tasted coffee and never will. It will never experience the warm rush of creamy bittersweet liquid flowing between its lips and over its tongue and down its throat; it will never feel the fabulous buzz of caffeine that brightens even the gloomiest outlook; it will never know the horrific daily letdown of lifting the mug that final time to discover there are no more sips to be had. Coffee (or whatever your favorite beverage may be) is a uniquely human experience, so if I ask a chatbot to write about it, all the bot can do is fake it.

Artificial Flavoring

I want to give you a taste of what I mean here, so I asked the chatbots kindly (because we have an odd tendency to be polite to robots), “Please write me a short poem about how amazing coffee is.” ChatGPT spit out this:

Given the same prompt, Gemini (Google’s chatbot) prepared an identical structure, as if two quatrains (four-line stanzas) with AABB rhyme scheme are the standard for “short poem.” And it even “copied” the same ideas, like ending with calling coffee a friend:

Some of these lines might sort of mimic my experience in a way that is kinda cute. But something crucial is missing that leaves me grimacing like bad coffee. No thanks.

I’ve tried getting bossier with the bots, suggesting they ditch the nursery-rhyme structure or make the language more contemporary or the imagery grittier or the literary devices more unexpected. They will apologize and ask for another try. None of their attempts get any closer to having what a bot can’t have—originality.

Real poems are made of raw experience and emotion, not predictive rhyming text. In fact, when writing a poem, I reject most every predictable word that comes to mind; what I crave are new combinations of word / sound / feel(ing) / spell(ing) that will have a reader think, Ooooh yes, THAT. It will tether us as humans to a shared familiarity in which we are less alone because another person has felt and described what we have also felt.

The need for that humanness is not limited to poems. In this essay, for example, I want every sentence to feel real—with words like kinda or ooooh—because I want to connect with you. So one big question to ask ourselves is this: If a chatbot by its very nature (haha) cannot have the essential qualities necessary to write like a real human, why and when and how would we want to use it?

What’s a Bot to Do?

Let’s be real. Not every type of writing needs to be poetic or personal or original or human-like. Even as snooty as I am about making my java jive,^[1] sometimes I settle for instant coffee granules and powdered milk (when I’m backpacking) or gas station dispenser cappuccino (when I’m road tripping and the coffee shops are closed). Context matters a lot.

Wanna know when I let the bots do the writing for me? One example is when I need a new rubric—those tables that divide assignment criteria into descriptions of full or partial or no credit. They’re valuable for clarifying the score breakdown, allowing students to confirm requirements and allowing me to grade faster and more fairly, akin to a measuring stick. To decide if I should make them with AI, my thinking went like this:

1. Creating a rubric is a non-personal writing task. It doesn’t need personality, I don’t need to claim credit for authoring it, and doing it myself wouldn’t improve any skills (at least not this far into my career, though student teachers might benefit).
2. Chatbots are great at making rubrics. I input my assignment description, and I’m impressed with the output (which I double check and adjust as needed). That frees me up for more of the human side of teaching, such as personal comments to students (which should definitely not come from a bot since students expect my opinion of their work).
3. Others in my profession would approve. We collect cool teaching ideas wherever we find them, and there is no expectation from administration, colleagues or students that the material we use is original. Having a bot generate lesson plans or assignments or rubrics would be acceptable.

Believe it or not, I’m demonstrating that in some specific cases you can take the shortcut. You just need to understand the parameters.

On the other hand, I can’t tell you how many hours I’ve poured into crafting and recrafting this essay, because here it matters that I present my very polished original thinking that continues to deepen the more I revisit the question of “What exactly am I trying to say?” My name is on this article—I care about owning the work and the words that go into it and adding my sincere opinions to this important human conversation.

But don’t stop there! Because even if you answered no to any of the questions above, it doesn’t mean you can’t allow a bot to help you with the writing process, as I definitely did with this article—depending on what policies you’re restricted by. Read on.

Chatbot Bans vs. AI Policies

I grew up in a religion that admonished against coffee, just like you might have gone to a school that prohibited chatbots. As a kid, I tasted mocha ice cream and became instantly obsessed with its one-of-a-kind flavor—but felt absurdly guilty because I didn’t know how broadly the no-coffee directive applied. For many years, I steered clear of anything labeled coffee / mocha / java / espresso or even tiramisu, just in case. And it’s possible you’re in a similar panic with AI—that you’ve been warned against it, maybe even under threat of expulsion, and could feel anxious even reading this article.

Let me explain my frustration with the all-out chatbot bans that happen in many school districts, where the concern over cheating has outweighed the rest of the conversation. To me, this is equivalent to abstinence-only education:

Not only are you likely to use it on your own, but you’re also likely to need to use it in the future. A 2024 article in Information Systems Education Journal argues, “Students should be empowered to use AI, and knowing when to do so is a critical first step. It is essential that they learn about these tools to be competitive in their fields of employment” (Firth). After all, technology advancements tend to be quickly integrated into professional settings, so you could enter the workforce lacking necessary AI skills if your education forced you to avoid it.

I’ve strained my coffee metaphor here, but I want to make this crucial point that there are ways AI can help you become a better student / writer / professional / person. And we’re gonna miss out if all we do is ban it.

Personally, I prefer to see thoughtfully incorporated AI policies that pinpoint what is acceptable to us as humans. In my classes, my policy is that cheating is cheating, whether you steal from a person or a robot:

If I suspect a student has cheated (generally because I can taste artificial flavor in the paper), we have a conversation about how to fix it in that assignment and avoid it in the future. The conversation itself often helps them understand their own humanness better since I point out the robotic writing that tipped me off. In one case it wasn’t robotic but simply didn’t sound like my student, and they confessed that an overzealous parent had edited the essay so heavily that it took on the dad’s style! The student and I laughed about pushy fathers, they agreed to rewrite it in their own voice, and they learned a valuable lesson about retaining autonomy of their work. (More about that coming up.)

I’m hopeful that describing such practices can persuade more educators and institutions to end all-out bans and allow AI in smart ways. You can likewise influence your teachers by writing convincing research papers on this topic if you feel strongly about it.

However, regardless of my opinion or yours, you’ll still need to bow to the rules of your institution, professors, employers, etc. Check their policies on appropriate AI usage before you dive in. But at the very least, you get to decide your own personal chatbot usage when it’s not for school. It might or might not be your cup of tea (or coffee) based on your values. It wasn’t until well into adulthood that I left religion and decided my own beverage preferences. As I share my coffee and AI viewpoints here, I’m hoping you are feeling out for yourself what’s best for you.

Chat to Learn and Learn to Chat (vs. Cheat)

“We should get coffee and catch up sometime!” is the best sort of invitation, isn’t it? It’s an offer of connection—that vital ingredient for healthy humanning.

Conversation is also something that draws us toward programs like ChatGPT. It’s literally in the name! In the dialogic exploration of questions and answers between two contributors, new discoveries unfold. When they’re both human, they learn more about each other—how the other person thinks or feels or lives or dreams—as well as whatever topics arise. When one is a bot and one is a human, both can still learn. The bot constantly updates its memory, learning more about what humans want and how we combine words; you can likewise uplevel your writing if you watch for what you can learn from the bot.

In fact, approaching AI interaction this way—to enhance your own thinking process—helps to combat legitimate objections to chatbot usage. As Julia McKenzie Munemo, the writing center director at Williams College, warns students directly,

We’ve come to see the goal of writing as getting to our point quickly, making a strong argument and concluding carefully, all with perfect grammar and syntax. But anyone who has revised a paper, come back to an idea after a sleep or a walk or a shower, or worked with a tutor to brainstorm new directions will tell you that the true goal of writing is to clarify, understand and experience our own thinking.

The Bot will steal that from you. If I could send one message to students, it would be: don’t let it.

I completely agree with her. Don’t let it do your thinking for you. Don’t let it steal the process of finding your clarity and determining what it is that you have to say. But what if we can learn to chat in ways that make us better thinkers?

In class, I have my chatbot-skittish students start with a very simple conversation:

The sample is very basic—just a 100-word reading response from their homework. But the emotion in the classroom becomes audible as students exclaim, “Wait, what? How does it know that about me from one paragraph?” And unlike the copycat nature of the chatbot poems, no two students get the same feedback. They get excited to keep talking to the bot. They pop in extra samples. They’re curious and engaged. The inconceivably enormous amount of writing the chatbot has analyzed means it has something ironic to offer us: enhanced connection to ourselves.

This new awareness of their own writing qualities leads my students to see more clearly what distinguishes them from a robot, so they lean into their real voice rather than trying to sound “smart” with words they don’t use IRL. Their essays sound more like them than before they chatted with the bot.

I’m not the only one valuing this type of dialogue. A 2023 article in Language Magazine asserts that gifting students more personalized feedback is one of the most positive aspects of using chatbots in education. Chatbots can benefit students with “immediate feedback on various areas of writing since [chatbots] process and respond to a text at near-lightning speed” (Baffour, emphasis added). No teacher or writing tutor is superhuman enough to compete, so we should be welcoming the help.

As a writing teacher, I wish could sit down with every student over coffee and talk about their writing. I type as much response as I can when I grade a hundred essays per batch, but here is a chance for a much greater quantity of personalized attention than they could possibly get from another human being. It’s equivalent to chatting with a professor-like tutor any time you want, and at any stage of the development process. As college-comp gurus Gerald Graff and Cathy Birkenstein say in their latest edition of They Say / I Say,

These sophisticated bots can also help you overcome that dreaded feeling of writer’s block, and keep you company during a long slog of writing, breaking through the isolation of the writing process and offering stimulating feedback that can sharpen your focus—even at 2 o’clock, if it suits you, in the morning. (299)

How ironic that they used phrases like “stimulating feedback” and “sharpen your focus”—just like a burst of caffeine when you’re feeling sluggish!

Customize Your Order

If you’ve ever gone to a coffee shop, you know how crazy specific the potential combinations of options can be. Not only do we have the basic menu choices like drip coffee, cappuccino, mocha, or latte, but we also tell them hot or iced, whole milk or almond milk or oat milk, 8oz or 12oz or 16oz, what sweetener, how foamy, and so much more.

With a chatbot, you get to—and need to—be just as specific.

Let’s say you’re working on an essay about the uses of chatbots in writing an essay. If that’s all you tell the chatbot, it might very well churn out an essay about the uses of chatbots in writing an essay, and you might suddenly panic because it just cheated even though you didn’t mean to ask it to do that! On the other hand, if you get as specific as a coffee order, you’ll automatically be in better shape:

When I fed this to Gemini, it gave me specific bullet points rather than an essay:

• Brainstorming and Idea Generation
• Research Assistance
• Writing Support
• Language Learning
• Peer Review and Feedback (“What are examples?”)

Each of these came with a detailed paragraph summing up how it could work. I then compared that to a human-written source: an article by Sabrina Ortiz entitled “5 Ways ChatGPT Can Help You Write Essays” that lists suggestions I’ve paraphrased here:

1. Generate topic ideas
2. Create an outline
3. Find potential sources
4. Make a sample essay as a guide
5. Catch editing errors

Research assistance / finding potential sources is on both lists, and it easily could’ve been the case that I learned of Ortiz’s article via Gemini. Great! Just make sure that you go to the source itself, confirm that it’s real (yeah, bots can sometimes “hallucinate” and make things up), and compare/contrast the source info to what the bot offered. In this situation, both lists are valid, just worded differently; in other situations, you might discover the chatbot told you something inaccurate.

Hopefully you can see the many advantages of using both chatbot- and human-generated content. Gemini was quick to offer a similar “opinion,” which I found amusing:

It’s important to remember that chatbots are a tool, and they should be used in conjunction with other resources and human interaction. Students should always take the initiative to research, think critically, and develop their own ideas. Using chatbots responsibly can enhance the writing process and help students produce better essays. (“What are examples?”)

Well said, Gemini! So I decided to quote it and cite it to give the chatbot credit—and to model MLA 9 chatbot citations for you. You’ll notice that—like all citations—what I put in the parentheses above matches the first few words of the full citation in my Works Cited section, allowing you to find it easily. To save you the trouble of jumping down there, here are the full citations for the Ortiz article and the “What are examples?” prompt:

Ortiz, Sabrina. “5 Ways ChatGPT Can Help You Write an Essay.” ZD Net.com. 19 Aug 2024. Web. 25 Oct 2024. https://www.zdnet.com/article/how-to-use-chatgpt-to-write-an-essay/

“What are examples of how chatbots can help with essays without helping a student cheat?” prompt. Gemini, 2.5 Flash version, Google, 25 Oct 2024, gemini.google.com/app

Notice how the chatbot citation is quite different in that the prompt is featured first rather than an author. Notice that the citation prompt is a summary of my much more complex prompt that I showed you in a box above. And notice how the chatbot’s name comes after the prompt. The prompt is considered more of an “author” than the bot!

Another way to think of specifying your request is giving the chatbot boundaries. For example, when my students are at the thesis statement stage of writing, I tell them to prompt the chatbot to help them strengthen their thesis by only asking questions. This way the chatbot doesn’t rewrite the thesis but instead helps guide the students toward deeper thinking that will lead to stronger ideas. Boundaries / specific requests help the chatbot understand its role in the conversation and let you retain autonomy of your writing.

For example, when I wanted feedback on this article, I fed it to ChatGPT with this detailed prompt:

It did just what I asked, and I was able to decide for myself how to incorporate its advice.

I also sought out selective human feedback—some of which highly differed from ChatGPT’s since humans have individual experiences and opinions that offer unique insights. By “seletive,” I mean I was choosy about my beta readers, anticipating whose perspectives might enhance my own thinking. The aspects that stood out to each bot/reader proved so helpful in strengthening the essay that I would not have wanted to rely on only AI or only humans. As Gemini said, chatbot and human resources “should be used in conjunction”—never as a complete robot takeover.

Oh and you might discover a cool side effect that the advice above about “customizing your order” improves human feedback as well (especially if you have an overzealous proofreader in your life who needs some boundaries). You’ll generally get better insights when the human understands your context, purpose, audience and request too. Just saying.

Get What You Want or Take It Back

Sometimes after you’ve ordered coffee, you take your lidded cup all the way to a cute little bistro table in the corner by a sunny window, hang your jacket on the back of the chair, settle in with your book or laptop or friend, take your first sip … and realize they accidentally gave you a dirty chai or an Americano or a macchiato or any other menu possibility that was not what you ordered.

Sure, you could drink it. Or … you could take it back to the counter and ask again for what you actually wanted.

This is a key step in talking with bots. Too often my students just accept what the bot gives them when it isn’t what they wanted! The student might feel like the bot knows better than they do. Nope. The bot is guessing what you want based on what thousands or millions of other users have asked for, but those other users aren’t you. Just because a pumpkin spice latte is what most customers are ordering in October doesn’t mean it’s what you were craving that morning.

Remember that this is your order—that you are the one giving the orders around here! As Graff and Birkenstein insist, “Students themselves need to take the lead, mentoring, directing, and often even correcting the chatbots themselves” (298).

• If the bot tries to take over and write or rewrite your paper for you, you can redirect it: “Actually that’s not what I’m looking for. Can you just give me suggestions for how to revise the writing myself?”

• If the bot spits out huge amounts of text when you wanted less, you can say so: “Thanks but that’s way too much at once. I’m overwhelmed by it. Can we just go one step at a time?”

Once you get the hang of this, it becomes easier to speak up—whether it’s a botched coffee order or a chatbot gone rogue. Learning to speak up is an important human skill that can enhance your life, like by raising your hand in class or meeting with your professor during office hours to clarify the thing you didn’t understand. So practicing with a bot is a good place to start.

Savor Your Humanness

There are far more ways to put chatbots to great use than I’ve listed here. In fact, the chatbots can tell you plenty more options much more efficiently than I can. Instead, my intention was to offer what the chatbots can’t: a human perspective on how to be even more human.

In that spirit, I wrote a free-form poem as a parting gift. It structured itself organically as I revised, creating a unique ten-stanza pattern I’m quite happy about. Likewise, the title is a new word I made up for the occasion. Each line felt delightful to pen from scratch—going from the wobbly toddling stage of early drafts to the crisp satisfaction of word combos coalescing.

For this, I used zero chatbot input because it sucks at poetry, and the first rule of seeking advice is to seek out someone skillful in that skill. Otherwise, it’s like asking my coffee machine to make toast! If you take nothing else from this article, remember to be choosy how and when you use AI, remember to ask it for things it’s good at, and remember to appreciate the originality that makes humans shine.

With love,

Nikki

Works Cited

Baffour, Perpetual. “AI Can Strengthen Student Writing, Not Weaken It.” Language Magazine. 25 Jan 2023. Web. 25 Oct 2024. https://www.languagemagazine.com/2023/01/25/ai-can-strengthen-student-writing-not-weaken-it/

Firth, David R., Mason Derendinger, and Jason Triche. “Cheating Better with ChatGPT: A Framework for Teaching Students When to Use ChatGPT and other Generative AI Bots.” Information Systems Education Journal 22.3, July 2024. pp. 47–60.

Graff, Gerald and Cathy Birkenstein. They Say / I Say. 6th ed, Norton, 2024.

Munemo, Julia McKenzie. “A Message to Students About ‘The Bot.’” Inside Higher Ed. 22 Jan 2023. Web. 25 Oct 2024. https://www.insidehighered.com/views/2023/01/23/message-students-about-%E2%80%98-bot%E2%80%99-opinion

Ortiz, Sabrina. “5 Ways ChatGPT Can Help You Write an Essay.” ZD Net.com. 19 Aug 2024. Web. 25 Oct 2024. https://www.zdnet.com/article/how-to-use-chatgpt-to-write-an-essay/

“Please write me a short poem about how amazing coffee is” prompt. ChatGPT, 2 Oct version, Open AI, 25 Oct 2024, chatgpt.com

“Please write me a short poem about how amazing coffee is” prompt. Gemini, 2.5 Flash version, Google, 25 Oct 2024, gemini.google.com/app

“What are examples of how chatbots can help with essays without helping a student cheat?” prompt. Gemini, 2.5 Flash version, Google, 25 Oct 2024, gemini.google.com/app

This essay is not from Writing Spaces. It is an original essay written for this Composition at CMU text to address chatbot usage, which was not covered in Writing Spaces volumes 1–6.