University-level writing is argumentative writing. You are not summarizing what other people have said — you are making a claim, supporting it with evidence, and engaging with views that disagree. Everything else is in service of that.
Where AI helps, where it hurts
AI is useful for pressure-testing a thesis, reviewing structure, catching gaps, and feedback on a draft you have already written. It is dangerous for generating the argument, producing prose in your voice, or drafting from scratch. The first set makes your writing better. The second set replaces it.
Thesis Strength · 20 XP
Which is the strongest thesis for a 10-page history paper on the New Deal?
A strong thesis is specific, arguable, and supportable. A reasonable reader could disagree — which is what makes it worth arguing. "Very important" and "both positive and negative" are statements no one would contest, which means there is nothing for the paper to do.
The draft-then-feedback loop
The highest-value use of AI in university writing is feedback on your own draft. You write the paper. Then you ask the AI for targeted critique — thesis clarity, evidence-to-argument fit, unaddressed counterarguments, the weakest paragraph. You take what is useful, ignore what isn't, and revise yourself.
The prompt that distinguishes this from "rewrite my paper" is an instruction the AI will respect: "Do not rewrite — advise me."
Prompt · 15 XP
The Four-Point Feedback Prompt
Here is my draft for [course, paper type]: [paste draft].
Give me feedback on four specific things:
1. Is my thesis clear and arguable? If not, why not?
2. Does my evidence actually support my argument, or are there gaps?
3. What counterarguments have I not addressed that a reader would raise?
4. Which paragraph is weakest, and specifically why?
Do not rewrite. Advise me.
Voice — why it matters
AI-generated prose has a consistent feel: smooth, evenly-paced, slightly over-formal, using a narrow band of sentence structures. Professors read hundreds of papers a semester. They can tell. A paper that reads like an AI wrote it — even if you personally wrote it — signals that you did not think hard enough about it to sound like yourself.
Your voice is built by the specific choices you make: which examples you reach for, where you pause, what you notice and what you don't, the rhythm you naturally write in. A draft you revised to sharpen those things reads better — to your professor and to future you — than anything a model produces.
Voice Spot · 15 XP
Which opening reads as a student's own voice?
The second opening has specific images (sewers, smell, cholera), an unusual claim (London built them "twice"), and a distinct rhythm. That's a voice. The others are interchangeable — the kind of prose that comes out of an AI's training distribution, or a student who hasn't found their way into the material yet. Your reader can tell.