Is It Cheating If GenAI Does My Writing for Me?
As my students learned, it depends.
Here are the key takeaways from the article from the TLDR bot:
AI Integration in Education: New tools and resources, like Open Scholar and MyEssayFeedback, as well as enhanced ChatGPT features, are being explored for teaching, research, and formative assessments.
Teaching AI Prompt Design: The author is teaching students how to use a structured Role/Task/Voice/Context framework for drafting prompts to generate AI-written paragraphs, aiming to enhance their rhetorical skills and critical thinking about AI use.
Ethics of Using AI in Writing: Classroom discussions revealed divided opinions on whether using AI for assignments is "cheating," with students' perspectives influenced by the depth of their prompts and acknowledgment of AI assistance.
AI as a Workplace Skill: The author emphasizes preparing students for professional environments where AI is increasingly expected as a tool, while cautioning that AI benefits skilled users more than novices.
A New Writing Genre: The rise of generative AI has introduced a unique genre with its own conventions, which educators and students must adapt to for effective collaboration and ethical use.
Created by ChatGPT 4o with Canvas.
Shiny New AI Toys
Before I get into this week’s topic, which is teaching your students to have ChatGPT straight up write a paragraph for them (yes, I went there), I wanted to share a few updates. It was another busy week in AI world. In case you missed it, here are a few things that caught my attention:
AI in Education: Leveraging ChatGPT for Teaching: It’s a new Coursera course (from Ethan Mollick)—free to audit, $29 certificate. Of course, I signed up.
Open Scholar from the University of Washington
A few weeks ago, I asked what truly open AI might look like, and now I have some ideas. Here's the new Open Scholar chat tool from University of Washington. It’s a RAG tool that is focused on computer science papers for now. My first query was not successful, but when I specified that I wanted references, I got good results. You can also ask questions about algorithms and concepts. They have detailed specs about their model development here. I’ve posted a screenshot of an interaction below so you can get a feel for what the UX looks like. I was satisfied with the results it returned to my query about articles on teaching first year writing with generative AI tools.
Chat GPT Student Writing Guide: Okay, let’s just say it. This is two years too late. When I attended the OpenAI Educator forum in October 2024, I was not super impressed with the company’s approach to AI in higher education. Now, in true OpenAI fashion, they’ve come up with a guide to writing. It’s mostly pretty good; I have tried many of these things in my own classes. But of course, they didn’t cite or acknowledge anyone for these ideas. I mean, the Internet is their oyster apparently.
Also, just in time for final papers (*sarcasm*), OpenAI has pushed writing updates for paid users of the 4o model. Here’s what they have to say about it:
“We’ve updated GPT-4o for ChatGPT users on all paid tiers. This update to GPT-4o includes improved writing capabilities that are now more natural, audience-aware, and tailored to improve relevance and readability. This model is also better at working with uploaded files, providing deeper insights and more thorough responses.”
Sure, but are you Claude? No, you are not.
Here’s a quick video I made that demonstrates the new writing style with the Canvas feature (but it still makes crappy art; see above).
Some Personal Fun News
MyEssayFeedback, a tool created by Eric Kean that I use for formative AI assessments in my English 102 and English 211 courses, is moving on to Phase II of the 2025 Tools Competition. With only 35% of 1000 applications selected to advance, this is really great news for Eric. I have thoroughly enjoyed piloting the tool over the past 18 months and providing feedback through its development.
Finally, I wanted to share some information about a course I am teaching at the College of Western Idaho this spring (I hope! It depends on how many people register). I’m simultaneously building it as an open (CC BY 4.0) course for Online Idaho.
ENGL 199 Writing in the Age of Artificial Intelligence
“English is the hottest new programming language.”—Andrej Karpathy, former OpenAI developer
How Will AI Impact Writing and Research at School, in the Workplace, and in Your Life?
Learn how to use AI for writing, or take your AI-use to the next level! ENGL 199 is a new course that introduces students to the evolving landscape of writing and research through the lens of generative AI tools. Students will explore how AI can assist in academic writing, research processes, professional writing, and creative expression, while critically examining the ethical implications and challenges that arise from the use of these tools. This three-credit, asynchronous online course will offer weekly optional live remote sessions. Taught by Liza Long, Associate Professor of English at CWI and one of two AI Fellows for the State of Idaho.
Okay, now onto this week’s lesson.
Is It Cheating to Have a Chatbot “Write” for You?
I decided to integrate generative AI in every aspect of my English 102 courses this semester. It’s been quite an experience, and I’m excited to have some course release in the spring to analyze data and report some findings (I got IRB approval to do content analysis on the students’ interactions with AI).
Throughout the course, students have been practicing using template prompts for various types of assignments. You can see some examples in the textbook I am writing, Cyborgs and Centaurs. All prompts are licensed CC0 (they’re in the purple boxes). I’ll be asking for more formal feedback from colleagues (maybe you???) next spring so that I can approve what I lovingly describe as a “hot mess.”
One of the final assignments in the course requires students to draft their own AI prompt to produce an example of writing in their workplace and test it on two different AI tools. For example, one nursing major asked ChatGPT and Google Gemini to create a natural birth plan, then analyzed the results (I’m sharing this with the student’s permission, and as an aside, make sure you check out the absolutely CREEPY AI images throughout her blog post).
To make sure their prompting skills were up to snuff, in class a few weeks ago, I asked students to use the Role/Task/Voice/Context approach to draft a prompt for their favorite chatbot, asking it to “write” a paragraph in support of their solution for their persuasive argument essay.
Here’s an overview of this approach if it’s new to you:
Role: This defines the persona the model should assume (e.g., nurse, teacher, expert in X, etc.)
Task: This specifies the action the model needs to perform, such as "write a natural birth plan for a patient.”
Voice: This describes the tone and level of formality the model should use, like "professional tone," or "friendly and enthusiastic” (that’s my own go-to voice in case you’re wondering).
Context: This provides background information or relevant details to ensure the model generates accurate and contextually appropriate responses.
By the way, while this level of specificity was important in early prompting, I am finding it much less necessary, especially with tools where I have a long history (and they have an increasingly long memory). But I like teaching this approach to prompting because it maps well to the rhetorical concepts we are learning in the class.
The exercise was interesting. All of the students were able to draft a prompt and generate a paragraph within the 10 minutes I allotted for this exercise.
Then I stood in the middle of the classroom and said, “Was that cheating?
Did I just ask you to cheat?”
Spoiler alert: I do not personally know the answer to this question. I was genuinely curious about their thoughts.
About one-third of my class thought it absolutely wasn’t cheating. One third thought it was definitely cheating. And the final third, like me, weren’t sure.
Then they asked me what I thought.
“I think that however you answered that question, you’re right,” I replied. “If you think you cheated, you did. If you think you didn’t cheat, you didn’t. And if you aren’t sure, you need to think about it more.”
We drilled down into the paragraphs the AI tools “wrote” to get a better understanding of why our class felt so differently about this exercise. It turns out that my students who did not think it was cheating had the most detailed prompts. In fact, their prompts (to produce a paragraph) were roughly equal in length to the paragraph the chatbot produced.
“Then what’s the point?” one of the students who thought using AI was cheating asked. “If you had to write a paragraph, why not just write your own?”
The first student explained that she felt more confident when she asked AI to help her organize her thoughts. The other students who thought it wasn’t cheating nodded vigorously. Then she added, “Also, I acknowledged and cited my AI use at the end of the paragraph.”
In my class, no matter what students do with AI, if they acknowledge, cite, and share their chats, I will not call it “cheating.” However, I do reserve the right to ask students to redo an assignment, often because the AI output they submit does not meet the assignment requirements.
My takeaway from this activity is that we need to pay attention to our own personal ethics and values when we interact with (or choose not to interact with) generative artificial intelligence. But I’m choosing to interact (and to teach students to interact) because I’m personally nervous about the workplace my students will enter. Based on my own experiences, genAI is most useful for people who already have expertise, not for novice writers. And if workplaces are already cutting entry level positions because AI can do their jobs, how will we train the next generation of experts? Can I train them to be experts now?
Existential threats aside (have I mentioned the need for universal basic income in the past five minutes?), in AI prompting, we now have a new genre, with its own rapidly evolving rhetorical conventions. I want to make sure that my students know how to interact with AI. Their workplace will expect it.
If you want proof, look no further than this email I got from our Center for Teaching and Learning, which, according to GPTZero (yes, I know, I know! These detectors are not accurate!), was 100% AI generated 😊. And no, I personally do not see this particular use of genAI as cheating. I would have done the same thing.
Happy prompting, and I sincerely hope that you get to enjoy at least a little time away for Thanksgiving! We all need to breathe.
P.S. This post was written entirely by me, a human, except for the key takeaways at the beginning, which were written by an artisanal custom GPT I handcrafted in my AI workshop while making homemade granola.