AI-Fluent, Not AI-Written: How I Tell Students to Use AI on Their Essays

Every student I work with is using AI. Not most of them — all of them, in one way or another. So the conversation in my office is no longer "should you use it?" That ship has sailed. The real conversation is: are you using it in a way that makes your essay better, or in a way that quietly erases you from it?

Because there's a difference between being AI-fluent and being AI-written. And readers can tell.

What AI-written actually looks like

Let me describe the essay I now see constantly. It's smooth. It's well-organized. It has a graceful opening image, three tidy supporting beats, and a conclusion that lands like a TED talk. It is also, somehow, about nothing. No specific detail survives for more than a sentence. The voice belongs to no one — it could be pasted onto any application, any student, any prompt.

Students think polish is the problem — that if they rough it up a little, add a contraction or two, it'll pass. That's not the issue. The issue is what's missing. A real seventeen-year-old's essay has texture: the oddly specific memory, the opinion that's slightly too strong, the sentence that only makes sense if you know the kid. AI produces the average of a million essays. Your essay's entire job is to be un-average in the specific direction of you.

And here's the part nobody warns students about: even if a reader can't prove anything, an essay that could be anyone's is worthless anyway. The essay is evidence — of how you think, what you notice, who you are. Outsourced thinking produces an empty exhibit. You haven't cheated the system. You've just shown up with nothing.

The line I draw

I tell students: AI can help you think. It cannot do your remembering.

The raw material of a good essay — the moment in the robotics lab, what your grandfather actually said, the thing you're embarrassed you believed at fourteen — lives in your memory and nowhere else. No tool can generate it, because it isn't information. It's your life. The moment you ask AI to supply the material, you've handed over the only part that was ever valuable.

But thinking? Organizing? Stress-testing? That's where fluency comes in, and students who refuse to use these tools on principle are leaving real help on the table.

How my students actually use it

As an interviewer, before drafting. The blank page is where essays go to die. I have students talk — literally talk — through their story while AI asks follow-up questions. "You said the summer job changed you. Changed how?" A good Socratic push surfaces details the student didn't know they had. The words in the essay will be theirs. The excavation can be assisted.

As a structural critic, after a real first draft. Paste in your own draft and ask: where does this lose energy? What's the actual claim of paragraph three? Does the ending earn the beginning? AI is genuinely good at this — pattern work on your text. What it writes back should be questions and observations, not replacement sentences.

As a fact-checker and researcher. For "why this school" essays, use it to find the professor, the lab, the program detail you'd otherwise spend an hour hunting. Then verify it on the school's actual site, because AI invents plausible-sounding programs with total confidence. Fluency includes knowing it hallucinates.

As a ruthless editor, at the very end. Word count over? Ask what's cuttable. Awkward sentence? Ask why it feels off — not for a rewrite, but for a diagnosis. The fix stays yours.

Notice what's not on this list: "write my essay about X." Or "make this sound better" applied to whole paragraphs — which really means "make this sound like everyone," and works exactly as advertised.

The two tests

Before any essay leaves my office, it passes two checks.

The read-aloud test. Read it out loud, in your own kitchen. If it doesn't sound like you talking — the way you'd actually explain this to someone — it isn't finished. Your cadence, your humor, your bluntness. AI's default register is a very competent stranger. Readers have met that stranger four hundred times this season.

The traceability test. Pick any sentence. Can you point to where it came from — a real memory, a real opinion, something you said in a session or wrote in a first draft? Every sentence should trace back to you. The moment you hit one that exists only because the tool generated it, it comes out.

A word on rules

Policies vary by school and they're shifting every cycle — some colleges permit AI assistance, some restrict it, some ask you to disclose. Read each school's actual policy before you submit, and follow it. There's no version of "but my consultant said" that rescues an application. When in doubt, the tests above keep you honest anyway: an essay built from your own material, with AI used as a tool rather than an author, is defensible almost everywhere — because it's genuinely yours.

The actual skill

Here's what I really think is going on. Universities aren't admitting students despite AI. They're admitting the first generation that will live its whole adult life alongside it. The skill they're quietly selecting for isn't avoiding the tool — it's having a mind of your own while using it. Judgment. Taste. A voice that survives contact with a machine that wants to average it away.

So yes, be fluent. Use every tool that helps you think more clearly. Just make sure that when a reader finishes your essay, they've met a person — and that the person is you.

Working on essays this fall and want help finding your angle — with or without the robots? Book a free 1-on-1 assessment and we'll talk through it.