Can You Use AI For Your College Essays? Has AI Made College Essays Sound Better? Well... Polished writing is NOT really what makes you memorable. Here’s what does.

Imagine your daughter shows you a rough paragraph about the Thursday tutoring sessions she kept running after the club's president graduated. One week later, nobody signed up. She thought the volunteers didn't care anymore... But when she asked one of them why, she learned that the late bus had actually been canceled! Now that she knew what the problem was, she moved the tutoring sessions to lunch, and now all 6 volunteers returned!

After reading it, you can see that the writing's not perfect. Uneven. But you can definitely see HER figuring out what went wrong as you read the story.

Then she pastes the paragraph on ChatGPT and, in the new version, she “transformed educational access” and “empowered the next generation!” Wow... quite a lot to accomplish by moving a Thursday tutoring session to lunch time!

The new essay surely has fewer awkward sentences. But where did the canceled bus context go? Or maybe... what happened to her wrong assumptions at the beginning and the follow up questions that helped her eventually figure out what the problem was?

And here is where “making the essay better” can accidentally make the essay worse... If polishing removes the mistaken assumptions your daughter started with, the weirdly specific questions she asked, the decisions she made along the way, or even the slightly imperfect way that she naturally explains something, you just end up with clean writing and considerably less of YOUR DAUGHTER in it.

This matters even more now that students are growing used to AI-generated writing. The more they read it, use it, and edit with it, the more familiar the vocabulary and sentence structures become. Over time, it's naturally difficult for a student who grew up like this to even notice when something sounds too polished and no longer carries the identity that it once did.

Parents are typically much better positioned to notice that. You've been listening to your daughter tell stories since she was a little girl, explain ideas, argue her cases, get excited about things, and, at times, take 12 minutes to explain something that should've taken 2. You know which expressions she actually uses when she speaks. You know what she notices when she's out on a walk. You know when an observation sounds exactly like something she would say at dinner, and when a beautifully constructed sentence makes you think: “She has literally never spoken like this in her life.”

Today, that makes you, as the parent of a high schooler growing up in the AI writing generation, more useful to the essay process than you may think.

Obviously, your job isn't to rewrite the essay in your voice or make sentences more sophisticated.

AI can help make writing more polished. But you can help make sure the person underneath the polish hasn't disappeared. So how do you do that? Let's get into it!

Starting the conversation with your child

Ask yourself: What did I just learn about my daughter?

Let's go back to our Thursday tutoring example. Imagine reading this sentence in your daughter's college essay:

“Leading the tutoring club strengthened my commitment to service and taught me the value of listening.”

It sounds nice. But... what did she actually listen to? Did someone challenge her plan? Did she initially misunderstand why nobody was showing up? What did she do differently once she realized she was wrong?

Now compare it with:

“When nobody signed up for Thursday, I thought our volunteers no longer cared... Then one of the volunteers told me the late bus had been canceled a week prior! Promptly, I moved sessions to lunch, and all 6 volunteers returned!”

Of course, these are hypothetical details, but the difference matters. The second version lets an admissions officer watch your daughter think through a problem she owned. She makes a mistaken assumption at first: that her peers had lost interest in the club. But then she identifies a different approach to solving this issue and, ultimately, asks a better question. She learns something she didn't know... And then she changes course because of it. We don't actually need to tell you that she “learned the value of listening” because you're now WATCHING your daughter listen, analyze what she learned, and then pivot.

And this is where parents can be REALLY useful. You know far more about your daughter's life than an admissions officer ever will. But you also have an advantage over her: she may not realize which parts of her own experiences are actually interesting because... well... she was there for ALL of them!

And this is where parents can be REALLY useful. You know far more about your daughter's life than an admissions officer ever will. But you also have an advantage over her: she may not realize which parts of her own experiences are actually interesting because... well... she was there for ALL of them!

We see this ALL the time in our work with students. But, importantly, this is not a “high school student” problem. Lara saw the exact same thing during her years as a management consultant at BCG, working alongside other Ivy League graduates and with senior executives in the C-suites of Fortune 100 companies.

So if your daughter spends 10 minutes explaining the logistics of the tutoring program and you immediately think, “Wait... go back. You thought the volunteers had stopped caring, found out you were wrong, and changed the entire plan once you understood what was ACTUALLY happening? THAT'S the interesting part.”... pay attention to that instinct!

Pay attention to your instincts!

You don't need to decide what the story means for her. But you CAN help her zoom out enough to notice the moments she has become too close to the story to recognize as important herself.

To your daughter, the canceled bus may feel like a completely irrelevant logistical detail. To you, it may be the detail that makes you stop and realize, “Wait... THAT'S when everything changed.”

Questions to ask your daughter:

• “What did you think was happening at first?”

• “How did you figure out you were wrong?”

• “What did you learn that you weren't expecting?”

• “What did you try first?”

• “Did your plan work immediately?”

• “What would the other people there say you actually did?”

And then... listen. Your daughter might remember a detail she'd never think of mentioning. Or... she might realize that something she had dismissed as completely ordinary is actually the part of the story that reveals the most about how she thinks.

But here's where you need to be careful: listen to HER interpretation before offering yours. If she tells you what happened and you immediately respond, “Oh! So what you learned is that good leaders listen to their teams!”... well, while that may very well be true, she might have taken something completely different from the experience. Not only that, but her interpretation might be more unexpected, creative, or uniquely HER than the lesson you saw from the outside.

That doesn't mean you shouldn't share what YOU noticed. You know your daughter incredibly well, and your perspective may help her see something about herself that she hadn't recognized before. Just don't let your interpretation become the protagonist of her essay. Share it, ask what she thinks, and if it genuinely resonates with her, it may add another layer to how she understands the experience. If it doesn't... let her surprise you.

Ultimately, your advantage as a parent isn't that you know how to write her essay better than she does. It's that you may notice the moments she has stopped noticing herself.

So instead of asking: “How can we make this sound more impressive?” Try asking: "What happened here that only YOU could tell me?"

You may actually be surprised by what she remembers once you give her the space to answer!

Research evidence on AI use

Okay... Can students use AI for college essays?
What does research actually tell us about AI and writing?

Why should parents worry about an essay becoming too smooth? Well, don't just trust us. There's already research that helps explain the risks. But we're going to be VERY precise about what it does and does not tell us, because none of these studies proves that an admissions officer will remember, admit, or reject a particular student because they used AI.

That said, it doesn't make the findings insignificant, though. Research doesn't need to prove that AI caused a particular admissions decision to tell us something useful. When we see meaningful patterns across hundreds, thousands, or tens of thousands of participants and applications, those patterns can be directionally informative about broader dynamics worth paying attention to, even when the studies cannot establish exactly what happened in any ONE specific student's application.

So we're not going to turn an association into a causal claim. But we're also not going to pretend the association tells us nothing. The interesting question is what these findings, taken together, might help us understand about what can happen to writing as AI becomes more involved in producing and refining essays.

Let's start outside college admissions. In a randomized experiment published in Science Advances in July 2024, Anil Doshi and Oliver Hauser assigned 293 people to write short fictional stories either with or without access to AI ideas. Then 600 evaluators, who did not know which writers had received AI help, rated the stories.

What they found was that writers with access to AI-generated ideas received higher individual ratings for creativity and writing quality. So... AI made the writing better? By those measures, yes. But there was yet another finding: the AI-assisted stories were also more similar to one another. (1)

Now, please keep the context in mind. These were short fictional stories, NOT college application essays. The evaluators were not admissions officers deciding whom to admit. But the experiment illustrates something VERY relevant to the example we started with: “This sounds better!” and “This sounds more like my daughter!” are 2 completely different judgments.

Interesting, though, right? Let's, now, look at what research is saying about AI in the college applications world. A February 2026 research by Jinsook Lee examined 81,663 undergraduate applications submitted between 2020 and 2024 to 1 selective university. Because the researchers obviously couldn't watch tens of thousands of applicants write their essays, they estimated AI use from patterns in the language.

That last detail matters for a reason that is a little more complicated than “AI hurt lower-income students more.” The researchers also found that estimated AI use increased disproportionately among fee-waiver recipients. Their interpretation was that AI may be taking the role, at least partially, of a substitute for the writing support these students wouldn't have access to otherwise. And intuitively... That makes a LOT of sense. A student who can't afford a private college counselor or hours of individualized editing suddenly has access to a tool that can review their writing, suggest revisions, and produce impressively polished language in seconds. (2)

But here's the uncomfortable part... Access to the same AI tool doesn't necessarily mean access to the same ability to use it the right way. A higher-income student might use AI along with a counselor, teacher, private advisor, or another educated adult who can say: “Nope. That sentence doesn't sound like you,” or “The original story was actually more interesting before ChatGPT rewrote it.” On the other hand, a student relying on AI precisely because they DON'T have the resources to that kind of human support may be asking the tool to play a much larger role in the process than the higher-income applicant. The Lead researcher in this study, Jinsook Lee, raised this as one possible explanation for the pattern, including differences in access to paid AI tools and knowledgeable human support.

Again, the study itself doesn't establish that this is what caused the difference. We can't look at this study and say that AI caused lower-income applicants to be rejected, or even that we know exactly WHY estimated AI use was associated with a larger decline in predicted admission rate for them. Remember: the researchers estimated AI use rather than observing it directly, fee-waiver status is only a proxy for socioeconomic status, and the data only came from 1 selective university. The authors themselves describe the findings as associations rather than proof of causation. Yet, that doesn't make the pattern irrelevant.

Directionally, it points to a much bigger question: What happens when the students who benefit the most from free writing support are also the most likely to let it replace their own voice?

And THAT brings us right back to you, parents.

You don't need to be a professional essay editor to provide something AI can't: years of context about what your daughter actually sounds like, notices, cares about, and thinks.

Finally, there is one more study we want to talk about. In September 2026, Calvin Isley, Johann Gaebler, and Sharad Goel studied 7,462 applications submitted between 2020 and 2025 to 1 public policy master's program where material AI use was prohibited. Their classifier estimated that 56% of applicants in the 2025 cycle submitted at least 1 essay primarily written by AI despite the college's published AI policies...

They also conducted a smaller exercise where 5 admissions officers rated essays. They found that AI was much better at improving grammar and clarity (as we would all expect) than at strengthening qualities like the writer's voice or evidence of service and leadership. Additionally, essays the admissions officers believed were AI-written received lower ratings. (3)

Again, context matters. These were graduate applicants, not undergraduate applicants. The study involved 1 master's program. A classifier can't tell us with certainty how every essay was actually written. But putting the findings together points us to a clearer conclusion: a beautifully edited and polished paragraph is not necessarily a revealing paragraph.

So... let's be realistic. Students are going to keep using AI. Some will use it thoughtfully. Some won't. And yes, some will use it even when a college explicitly tells them NOT to. We don't think pretending otherwise is helpful, so here we are... Facing the reality. However, “everyone is using it” is also a pretty terrible reason for your daughter to stop thinking critically about HOW she uses it. If anything, learning where AI helps and where it starts taking over matters even more now.

Understanding AI policy on college applications

Help your daughter check the AI policy at EVERY college she's applying for

Now... for a complication that can catch some people by surprise... Colleges do not all describe acceptable AI use in the same way. That means the exact same use of ChatGPT could be acceptable under one university's published policy but prohibited under another's.

For example, Yale's undergraduate admissions policy allows AI for general topic suggestions and grammar or spelling review. However, Yale says that submitting substantive AI output as the applicant's own work constitutes application fraud. Brown's undergraduate policy is narrower: it prohibits AI use in preparing application content, with an exception for basic spelling and grammar proofreading. (4, 5)

So a brainstorming request that may fit within Yale's published guidance should NOT automatically be used for a Brown essay. This is somewhere parents can be particularly helpful because, by application #9, there are a LOT of small things to keep track of. We would suggest adding the college's AI policy link, the date your daughter checked it, and the type of AI assistance she intends to use to the application spreadsheet from our supplemental essay guide (linking it here if you have not yet had a chance to read it).

Sure, you can help her find the policy. But let HER read it and decide how to work within it. And if the university's wording doesn't answer your questions, encourage her to contact the admissions office directly.

There is yet another problem here that has nothing to do with an essay “sounding like AI,” and we see it particularly often in the first drafts of the “Why This University?” essays. If ChatGPT provided the reasons your daughter supposedly wants to attend a particular university... does she actually know if those reasons are true? Has she actually researched those claims enough to know she agrees with them? And what might ChatGPT have NOT mentioned that your daughter, after doing her own research, finds genuinely exciting or relevant to why SHE wants to attend that university?

And even if they ARE true, there is another problem: AI gravitates toward certain reasons, words, and phrases over and over again. We're sure you have probably seen enough AI-generated writing by now to recognize some of the usual suspects: delve, embark, tapestry, landscape, ever-evolving, thought-provoking, utmost, transformative, leverage, streamline, turn of events, stark reminder...

The problem here is that ChatGPT is probably suggesting those exact same reasons to another thousand students applying to the same university as your daughter.

So now we have TWO problems:

1. The reasons AI generated for why YOUR daughter wants to attend that university are probably not the reasons SHE would have chosen at all. And even if some are, what else might she have discovered if she'd researched the university herself BEFORE asking ChatGPT?

2. The university may be reading essay after essay praising that exact same “interdisciplinary research culture,” or whatever other reasons AI happens to be favoring that year.

And THIS is how a technically accurate answer becomes a VERY generic one...

The goal of the “Why This University?” essay isn't to find the most impressive thing you can say about the university. It's to find the specific intersections between what that university offers and what actually matters to YOUR daughter. (Oh, and if you want to understand more about how we specifically recommend students approach the “Why This University?” essay, start with Part 1 of our 2-part “The Research University Trap” series by clicking here).

As we discuss in our 2-part “The Research University Trap” series, that impressive research program AI found may not accept undergraduates after all. Or the opportunity may have eligibility requirements your daughter doesn't meet. Or it may be perfectly accessible and genuinely excellent... but have absolutely nothing to do with why that particular opportunity makes sense for HER.

This is exactly why we recommend researching the university BEFORE asking AI to turn that research into beautiful sentences.

Anyhow, the takeaway here is: Follow each college's actual policy and have your daughter keep her notes, outlines, and earlier drafts. No, not because you're gathering evidence to outsmart AI detectors. Keep them because those drafts WILL help your daughter return to what she originally wanted to say before those 14 rounds of AI editing have made that almost impossible to remember.

Making sure your child doesn't lose their voice

Be the person who asks.
... Not the person who rewrites.

Concluding now: what can you actually DO when your daughter hands you a college essay?

1. Ask her to highlight every sentence that describes something she personally noticed It could be a choice she made, an idea she changed, or a result she actually observed. Then look at what's left. If almost nothing is highlighted, wait. The problem may not be the writing YET.

2. Read ONLY those highlighted sentences Ask yourself: could another student from the same tutoring club have written all of them? If yes, keep asking: Which options did your daughter consider? Did she misunderstand anything? If so, what? If not, did everything coincidently just fall into place naturally? Did anyone disagree with her? If so, about what? What happened after her first plan failed? How many plans did she try until she reached the final resolution? What did SHE notice in the process? What would she have done differently knowing what she know now?

No, you're not trying to INVENT uniqueness. You're trying to dig what's actually valuable and that might have been there in earlier drafts before the writing became generic.

3. Ask your daughter to read the essay out loud (this is our favorite trick!)
   And if she gets halfway through some beautifully constructed sentence and one (or both) or you notice that she would never say that... Ask her what she WOULD say instead. No overthinking, no polish. That's what she should be starting with.

We know why this is hard. You want to give your daughter every possible advantage, and when SO MUCH about college admissions feels uncertain, a beautifully polished paragraph feels understandably reassuring... It feels sophisticated, looks finished. All those awkward sentences disappeared.

But you have also seen the less polished version of your daughter, sitting in the corner of the kitchen table. The one who gets things wrong and worries about it. The one who always asks one-too-many question. The one who never gives up and keeps trying.

An admissions officer may never meet that version of your daughter... So yes, help her make the writing better. Just make sure that somewhere along the way, you don't edit HER out of it.

Sources

1. Doshi and Hauser, “Generative AI Enhances Individual Creativity but Reduces the Collective Diversity of Novel Content,” Science Advances, July 12, 2024

2. Lee and colleagues, “The Digital Divide in Generative AI: Evidence from Large Language Model Use in College Admissions Essays,” research preprint, February 2026

3. Isley, Gaebler, and Goel, “AI-written admissions essays are widespread but penalized,” research preprint, September 2026

4. Yale Undergraduate Admissions, “AI Policy”

5. Brown University Undergraduate Admission, “Integrity in the Application Process”

6. College Board Research, “Policy and Practice in U.S. College Admissions: Insights from a 2025 Survey of Admissions Professionals,” April 2026