Picture a student, named Jordan. He’s just used AI to get through a rough draft of a paper, and when you ask him what he actually thinks about the topic, he goes quiet for a second. “I don’t really know what I think about this,” he admits. “I know what it thinks.”
That admission isn’t really about cheating. Jordan hasn’t broken a rule. He’s used a tool that’s sitting right there, the way students have always reached for whatever tool was sitting right there.
Calculators were going to ruin math. Spell-check was going to ruin writing. Every new tool gets the same panic, and underneath it is always the same fear: that leaning on something outside yourself will stop you from ever standing on your own.
It’s a fair fear. But watch how selectively we apply it.
Nobody thinks less of a memoir for having a ghostwriter behind it, or a novelist for working with an editor who reshaped half the manuscript. Notice what’s different about those two kinds of help, though. The difference is the whole question.
A ghostwriter produces the words for you. Hand over the story, get back a finished product, and your own writing ability is exactly where it started. An editor does something else. Hand over a draft, get back questions and cuts, and if you engage with those notes, you come out sharper than you went in. Both are outside help. Only one of them builds something in you.
That’s the line worth drawing around Jordan’s situation, and it’s the line worth drawing around how you use these tools too. Ask AI to write the essay, and you’ve hired a ghostwriter. Ask it to poke holes in your argument, or explain why the first draft wasn’t landing, or quiz you until an idea actually sticks, and you’ve hired an editor. The tool doesn’t decide which one it is. You do, by what you ask of it, and by whether you let it leave anything behind in you.
I made a version of this case in the book: that asking for help is a strategic move, not a failure of independence. I meant that mostly about office hours and tutoring, the kind of help you could see and thank in person. AI complicates it, because the help is instant, tireless, and endlessly patient in a way no mentor ever is. That makes it easy to lean on constantly, without ever quite noticing you’ve stopped leaning on anyone human at all.
Here’s what worries me more than whether a student like Jordan borrows a paragraph he should have written himself. It’s whether he’s slowly trading away the people who would have helped him write it.
Think about the last time you solved a hard problem, not a homework problem, a real one. Did you work it out alone in a room? Or did you text a friend, or sit with a mentor’s old advice still shaping how you approached it, or argue it out with someone until the answer got clearer for both of you?
Most real thinking happens in the company of other people. That’s not a flaw in how we learn. It’s the whole design. The study group, the office hour, the late-night conversation that reframes a problem you’d been stuck on for days — none of that is an inefficiency waiting to be optimized away.
AI is sitting right there at one in the morning, when your friend is asleep and office hours are three days off. That part is genuinely useful. It’s also exactly why it needs to be used on purpose. Every time you reach for it instead of a person, you’re not just changing how you get an answer. You’re trading a human relationship for a frictionless one, and frictionless isn’t always the better trade. A lot of the trust in this life gets built in exactly the friction of asking someone real.
So here’s the question I’d want Jordan to sit with, staring at a blank document with a chat window open. Not just is this going to make me sharper, or just finish the job for me — he already half-knows that one. The harder question is the one underneath it. Is this actually the moment for a machine? Or is it the moment you’re avoiding a person, because the person is slower, and harder, and might ask you something back?