The Right and Wrong Uses of AI for Secondaries
It’s that time of the year again: secondary application season.
And here’s one major pressing question: when the admissions readers open your secondary essays, are they going to feel inspired and impressed?
Hopefully. But if you’re not strategic, they’ll just finish them, forget them, and move on. Because they’ve already read something similar, multiple times, maybe even in a single sitting.
The main challenge is simultaneously handling the huge volume of secondary essays, while still doing so in a way that stands out.
At first glance, AI might seem like a tempting shortcut for this problem, but it’s not that simple. AI is definitely not a catchall solution or cheat code for your secondaries, but it can help and save you time if used wisely.
The applicants who use it well are invisible, but the ones who are using it wrong are very obvious. Here’s what you need to know to make the best choices:
WHAT AI IS ACTUALLY GOOD FOR
Research: Feed it a school’s mission statement and ask it to identify the two or three core themes. Then go verify it yourself. You’re still the one deciding what’s relevant and whether it actually connects to your background. That’s time well spent.
Brainstorming: Stuck on a prompt? Ask AI what kinds of experiences typically address it, then use that list to figure out which of yours actually fits. You’re using it to think, not write the essay for you. That’s a key difference.
Structure: If you’re staring at a blank page, asking AI for two possible ways to organize your response is fine. Pick one. Then write the actual essay yourself.
WHAT IT LOOKS LIKE WHEN IT GOES WRONG
Here’s a sentence I have read, in some variation, hundreds of times this year:
"[School Name]’s commitment to community-based primary care deeply resonates with my experiences in underserved clinical settings, where I developed both technical competencies and a profound appreciation for health equity. I am confident that [School Name]’s innovative curriculum will allow me to build on these foundations while cultivating the skills necessary to serve diverse patient populations."
That essay was written by AI. I know that because every clause in it is technically correct and none of them say anything. “Deep resonance.” “Profound appreciation.” “Innovative curriculum.” These phrases exist to fill the shape of a secondary essay without doing any of the work.
Swap in any school name. It still makes sense. That’s unacceptable.
AI writes in an averaged voice, the aggregate of every application essay it’s been trained on. The output is fluent and organized, but sounds like nobody. It names your clinical experience without saying anything real about what you saw there. It mentions the school’s rural health track without connecting it to anything specific about your life. It sounds like a secondary essay the same way a stock photo looks like a real place.
Admissions committees have been reading AI-assisted secondaries for multiple cycles at this point. They don’t need detection software. The writing announces itself.
THE FOUR TELLS
These are the patterns that come up most when I talk to admissions professionals about what’s changed in the last two cycles.
1. Generic specificity.
The essay names real things: actual programs, actual faculty, actual community partnerships. But it doesn’t connect them to anything personal. It reads like someone asked AI to make the essay sound researched without making the research personally relevant to the writer. Admissions readers at schools with distinctive missions spot this immediately because they know what genuine engagement with their program actually looks like.
2. The balanced hedge.
AI loves false balance. “While my clinical experience has grounded me in patient care, my research background has shown me the importance of evidence-based medicine.” Read that again. Both halves say nothing. Real people don’t write like that. They write with a point of view, and sometimes that means one side of the sentence is stronger, more important, or more relevant than the other.
3. Assembly-line transitions.
Human writers get stuck. They backtrack. They make connections that are slightly imperfect. AI produces prose where every paragraph flows into the next with professional ease, and the result reads like it was produced rather than written. Readers notice the absence of any friction.
4. No moment.
A real secondary essay has at least one specific moment: a patient, a conversation, something you noticed that only you could have noticed. AI essays have impressions instead of moments. “During my time in the clinic, I witnessed the importance of patient communication.” That could describe anyone’s experience or observation.
THE POLISH TRAP
The most expensive mistake isn’t asking AI to write the essay from scratch. It’s asking it to improve one you already wrote.
Polish works like bleach on voice. Every slightly unusual word choice, every sentence that runs long because you were thinking, every rough transition that reflects how you actually connect ideas: AI smooths all of it out. The result is cleaner and significantly less yours. Admissions readers who spend months inside secondary essays develop a feel for the difference between a sentence someone chose and a sentence that got produced.
If you want to use AI on a draft you’ve already written, use it to catch typos. That’s it. The moment you ask it to “make this better,” you’re asking it to make it less you.
Use AI before you write. Don’t use it to write. And definitely don’t use it after you’ve written something real.
The applicants I’ve seen use it well this cycle are the ones who close the ChatGPT tab once they’ve figured out what they want to say. The essay that comes out of that process is specific, slightly imperfect, and sounds like a person. That’s exactly what it’s supposed to sound like.