Case Study 2: Priya's Essay — AI Assistance or Academic Dishonesty?

The Setup

Priya Ramirez is a second-year undergraduate majoring in environmental science. She's taking a general education course, "Technology and Society," that requires a 2,500-word research essay on a topic of the student's choice within the course's scope. Priya has chosen to write about carbon capture technology — a topic she's genuinely interested in from her major coursework.

Priya is a strong student, but she's juggling a heavy course load, a part-time campus job, and a family situation that's been demanding more of her time lately. She's not looking for a shortcut — she cares about learning. But she's also exhausted and behind schedule, and she knows that AI tools exist that could help.

The course syllabus includes an AI use policy that reads:

"Students may use AI tools (such as ChatGPT or similar) as research and learning aids. AI may be used for brainstorming, outlining, and exploring ideas. However, the final submitted essay must represent the student's own analysis, argumentation, and writing. Substantial passages generated by AI and submitted as the student's own work constitute academic dishonesty. When in doubt, disclose your AI use."

Priya reads this policy and thinks she understands it. But as she begins working, she discovers that the line between "research and learning aid" and "doing the writing for her" is blurrier than the policy suggests.

Priya's Process

Here's what Priya actually did, step by step:

Step 1: Topic Exploration (AI use: moderate)

Priya asked an LLM: "What are the main arguments for and against carbon capture technology as a climate strategy?" The AI provided a structured overview with six arguments on each side. Priya read through them, recognized some from her coursework, found a few she hadn't considered, and used the list to narrow her essay's focus.

She also asked: "What are the most important peer-reviewed studies on carbon capture effectiveness?" The AI provided a list of eight studies with authors, titles, and brief descriptions. When Priya searched for these in her university's library database, she found that five were real and three appeared to be fabricated — titles and author names that didn't correspond to actual publications. She used the five real ones and found three additional studies through the library's own search tools.

Priya's reflection: "The brainstorming was super helpful — it was like talking to a really knowledgeable friend who could organize ideas clearly. But the fake citations were scary. If I hadn't checked, I would have cited papers that don't exist."

Step 2: Outlining (AI use: moderate)

Priya wrote a thesis statement herself: "While carbon capture technology shows promise as one component of climate strategy, its current limitations — cost, scalability, and the moral hazard of reducing urgency for emissions reduction — mean it should complement, not replace, aggressive decarbonization."

She then asked the AI to suggest an essay structure for this thesis. The AI proposed a five-section structure. Priya modified it, combining two sections and adding one that the AI hadn't suggested (a section on environmental justice implications, drawing from a lecture in her major).

Step 3: Drafting (AI use: varied by section)

This is where things got complicated.

Section 1 (Introduction): Priya wrote this entirely herself. She wanted her own voice to set the tone.

Section 2 (How carbon capture works): Priya wrote a rough draft explaining the technology, then asked the AI: "Can you make this explanation clearer for a non-technical audience?" The AI returned a revised version. Priya used some of the AI's phrasing and added back technical details she thought the AI had oversimplified. The final section was a hybrid — roughly 40% her original words, 40% AI-revised phrasing, and 20% new material she added.

Section 3 (Arguments for carbon capture): Priya asked the AI to summarize the main evidence supporting carbon capture. She read the AI's summary, then wrote her own version based on the actual studies she'd found in the library. The final section was almost entirely in her own words, but her argument structure was influenced by the AI's summary.

Section 4 (Limitations and counterarguments): Priya wrote this section herself, drawing on class discussions and her own reading. She didn't use AI for this section at all.

Section 5 (Environmental justice): Priya wrote this entirely herself. It was the section she was most passionate about and where she drew on knowledge from her major that the AI wouldn't have connected to the essay's topic.

Conclusion: Priya drafted a conclusion, then asked the AI for feedback: "Does my conclusion effectively tie together my arguments?" The AI suggested adding a forward-looking sentence about policy implications, which Priya liked and wrote herself in her own words.

Step 4: Revision (AI use: light)

Priya pasted her complete draft and asked the AI to identify "any logical gaps, unclear transitions, or unsupported claims." The AI flagged two paragraphs where it said her reasoning was "underspecified" and suggested she add supporting evidence. Priya agreed with one suggestion and added a citation. She disagreed with the other and left her original argument.

She also ran the essay through a conventional grammar checker (not AI-based) for typos and mechanics.

Step 5: AI Disclosure

Priya added a note at the end of her essay: "I used ChatGPT to brainstorm ideas, check my outline structure, improve clarity in one section, and review my draft for logical gaps. All arguments, analysis, and research verification are my own."

The Questions

Priya submitted her essay feeling mostly good about it. She'd done real research, real thinking, and real writing. The AI had been a tool — like a very smart study partner who was available at midnight.

But she also felt uncertain. Had she crossed a line? Specifically:

Question 1: The Hybrid Section

Section 2 was roughly 40% AI-revised language. Priya had written the initial draft, but the AI had restructured sentences, replaced words, and smoothed transitions. The final version was better than what Priya would have written alone — but was it "her" writing?

On one hand: Priya provided the ideas, the technical content, and the initial structure. The AI improved the expression. This is not fundamentally different from having a writing center tutor revise her phrasing — except the AI did it faster and more extensively.

On the other hand: A writing center tutor teaches Priya to improve her own writing. The AI did the improving for her. Priya didn't learn how to write more clearly; she learned how to get an AI to make her writing clearer. These are different skills.

Question 2: The Research Shortcuts

Priya used the AI's initial list of arguments to structure her essay's scope. Even though she ultimately did her own research, the AI's overview shaped what she looked for and how she organized it. Did the AI do the intellectual work of "figuring out the landscape" that the essay assignment was designed to teach?

On one hand: Researchers routinely consult review articles and encyclopedias to get an overview before diving into primary sources. The AI served a similar function.

On the other hand: The process of struggling through the landscape — reading widely, encountering dead ends, gradually building a mental map — is part of the learning. By getting a neat summary from the AI, Priya bypassed a struggle that had educational value.

Question 3: The Feedback Loop

When Priya asked the AI to review her draft for logical gaps, the AI effectively served as a writing coach. Is this different from asking a human reader for feedback?

On one hand: Getting feedback from readers is a standard and encouraged part of the writing process. The AI provided the same service.

On the other hand: A human reader brings their own understanding and perspective. The AI's "feedback" is generated by the same pattern-matching process that generates its writing — it's not evaluating Priya's argument the way a critical reader would. It's identifying patterns that look like "logical gaps" based on its training data.

Question 4: Equity

Priya has access to AI tools because she has a smartphone and knows how to use them. Some of her classmates don't use AI at all — some by choice (they view it as cheating), some by unfamiliarity (they've never tried it), and some by principled opposition (they believe in doing their own work entirely). Priya's essay may be stronger than it would have been without AI assistance.

Is this fair? Is it different from the advantage that a student has if their parent is a professor in the relevant field and offers feedback on the essay? Or the advantage of attending a university with a well-funded writing center?

What the Professor Thought

Priya's professor, Dr. Okonkwo, read the essay and found it strong — well-structured, clearly argued, and well-researched. The AI disclosure was appreciated but raised questions he wasn't sure how to answer.

"Priya's essay demonstrates understanding of the material and genuine critical thinking," he noted. "Her environmental justice section is the strongest part — it's original, it draws on knowledge from outside this course, and it shows real intellectual engagement. But I can also tell that some sections are smoother than her usual writing. The disclosure tells me she used AI, but it doesn't tell me how much of the final text is AI-generated versus AI-revised versus entirely her own."

Dr. Okonkwo gave Priya an A-minus. He deducted slightly because he felt that Section 2's clarity, while impressive, didn't reflect Priya's demonstrated writing level from earlier assignments — suggesting the AI's contribution to that section was more substantial than "improving clarity."

He also began rethinking his AI policy. "The policy says AI can be used for brainstorming and outlining but the final essay must be the student's own work. Priya followed the spirit of this policy, as far as I can tell. But 'the student's own work' is becoming an increasingly complicated concept."

The Bigger Picture

Priya's case is not unusual. Across higher education, students and professors are navigating the same questions with inconsistent guidance and genuine uncertainty. Different institutions, departments, and individual instructors have adopted policies ranging from total bans on AI use to enthusiastic encouragement of AI tools as part of the learning process.

The stakes are real. Students who use AI extensively may develop weaker writing and research skills — or they may develop new skills (prompt engineering, AI-assisted research, human-AI collaboration) that are genuinely valuable. Students who refuse to use AI may develop stronger independent skills — or they may fall behind peers who are learning to leverage these tools.

And underneath all of this is a question about what education is for. If the goal is to produce a polished essay, AI makes students more efficient. If the goal is to develop critical thinking and communication skills through the process of writing, AI may undermine the purpose. If the goal is to prepare students for a world where AI tools are ubiquitous, avoiding AI in education may be counterproductive.

Discussion Questions

1. Based on the course syllabus AI policy and Priya's actual process, did she commit academic dishonesty? Where would you draw the line?

2. Is there a meaningful difference between Priya asking an AI to improve her writing and Priya asking a talented friend to rewrite a paragraph for her? If so, what's the difference?

3. Dr. Okonkwo noticed that Section 2 was "smoother than her usual writing." Should professors be grading based on the quality of the final product, or on whether the product represents the student's unaided abilities? What are the implications of each approach?

4. Write an AI use policy for a college course that you think is both clear and fair. What would you allow? What would you prohibit? How would you handle the gray areas?

5. Priya's environmental justice section — the one she wrote entirely herself — was the strongest part of the essay. What does this suggest about the relationship between personal knowledge, genuine interest, and the quality of academic writing?

6. The equity question: Is Priya's AI use different from other forms of advantage (tutoring, parental expertise, writing center access)? If so, how? If not, should all these advantages be treated the same way?

Connection to Your AI Audit Report

If the AI system you're auditing is used in educational contexts (or could be), consider: - How are students currently using this system? Is that use consistent with learning goals? - What policies govern its use, and are those policies clear, enforceable, and fair? - Does the system create equity issues between students with and without access or familiarity? - What skills might students gain or lose through interaction with this system? - How should assessment practices change to account for the system's existence?