Should you be using AI at all, in any way, for your college application writing? Not really.
Your application has many pieces: your transcript, which is just a list of classes and grades; a list of your activities, maybe with short descriptions; possibly some test scores; maybe some letters where teachers and counselors can talk about you. It’s all pretty impersonal, and the writing portions are your clearest and best possibilities to show colleges who you are as a person. So it’s completely counter-productive, counter-intuitive, short-sighted and just plain nuts to outsource that task to a literal robot! It’s one of the most desperate and self-sabotaging things you can do with your admission. Don’t do it.
(And spend some time figuring out why you were tempted in the first place. Is it a problem with confidence? With time management? Are you too caught up in social pressure to get into the “best” school that you become wiling to take shortcuts and make bad decisions if you think it will help you get accepted? Do you cynically believe that everyone else is doing it and that it’s the only way to keep up? If you’re letting AI write your entire essay for you, there’s a problem. It’s worth looking into.)
But you probably—hopefully—aren’t actually tempted to let AI write your entire response for you. If you are tempted to use AI, it’s probably as a tool to help you write, not a tool to write for you. There’s a difference.
I’ve used ChatGPT to help me with a few blog posts, just to experiment and try it out. My own writing process is very segmented. I begin with a fairly detailed outline, and I usually let days or even weeks pass before actually writing it out. The writing itself is mostly just typing—there’s very little though involved. And then I spend a lot of time revising and editing. So I’ve entered my outline into ChatGPT and asked it to do the typing for me. My outline and prompting are detailed enough that even the first draft of the AI-generated text only comes up as 10-15% likely to be written by AI when I run it through ZeroGPT. And then I go through the same revising process I normally do. The few times I’ve done this it actually took me more time and attention than usual. ChatGPT doesn’t write the same way I do, so the editing takes much longer. But it was interesting to try it out, and I don’t feel like I “cheated” in any way. I did all the thinking, and by the time I finished editing I even did most the writing. If you also keep your pre-writing, drafting, and revsision completely separate, and if you don’t mind spending more time and energy, then you may also want to use ChatGPT the way I have. You probably don’t.
The way I’m thinking right now about how a student can ethically, efficiently, and effectively use AI is to treat it as a person—a knowledgable but dull person. (I’m a knowledgeable but dull person, so no offense to AI.) Only use AI in a way you would use some knowledgeable but dull person. When I made the shift from being a public high school English teacher to being a college admission consultant, I gave myself a rule for helping clients with writing: I don’t do anything as a paid consultant I wouldn’t do as an English teacher. That’s the same advice I’d give a student: don’t let AI do anything your English teacher wouldn’t do.
Brainstorming. I love helping students think about topics to write about for their main Common App essay. I especially love the moments—they’re surprisingly common—when a student tells me a story, I say “that would make a great essay!” and the student says they wouldn’t have thought of it without our conversation. I did that sometimes during lunchtime or after-school discussions with students when I was a teacher, and I do it now with all my clients. This is a fine use for AI. I gave ChatGPT the prompt “I want to write my college admission essay. What are some good questions I should ask myself or exercises I should do to brainstorm good topics that may not be obvious?” It gave some pretty good starting advice, even if I’m skeptical that current Large Language Model AI can give advice that’s not obvious.
Giving commentary and suggestions. As a teacher I gave tons of suggestions to students on their drafts, and that’s now one of my primary jobs. I never rewrite anything, and I never insist that my suggestions must be followed, but I try to give thoughtful comments. This is something else that AI can do very quickly. I tried this out by inputting a never-followed-up-on first draft from years ago. And I told ChatGPT: “Here's a prompt and first draft for an application essay. What can I do to make it stronger? I'm trying to emphasize my multi-cultural background, my attention to detail, and a willingness to talk about things outside academics. Don't rewrite anything, just give me hints about what to focus on for my next draft.” The response I got was much better than I expected. ChatGPT didn’t give any feedback I wouldn’t have come up with on my own, but it put it succinctly. And it was instantaneous, which a live person can’t do. If you’re doing something similar, it’s really important that you spell out what it is you’re trying to make clear. You can’t just ask for ways to “make it better,” or “make it stand out.” That’s too vague and will only get you vague responses. I always read student writing with a focus on what qualities they’re trying to highlight, because by the time they’re writing essays we’ve already discussed that a lot. AI can’t do this unless you specify for it. It’s also extremely important, with AI or a real-life person, to explicitly say that you don’t want them to write anything, just to ask questions and give suggestions.
Give suggestions for grammar, punctuation, and style. This is the simplest and most obvious place to let AI give you a hand. It’s not significantly different from the spelling and grammar checking that most software has been doing for decades. So all I’ll say is to make sure that when you use AI for this, you make sure to only get suggestions, not to let it rewrite anything. And the more complex or vague the issue, the less confident you should feel in its suggestions. Simple comma or semicolon problems, no big deal. But when it starts suggesting ways to change sentence structure and tone, the more likely it’s trying to make you sound like an AI program and not like you. Watch out for that!
Recommend what to cut out to get it down to appropriate length. If you have a response you’re happy with, but it’s longer than the maximum word count allowed, whether for a 650-word essay or a 100-word supplemental response, go ahead and let AI recommend some ways to cut it down to size. I do this all the time for clients, and I used to do it a lot for students as a teacher. But again, don’t let it rewrite anything, just ask for suggestions that you can decide to incorporate or reject. And never, ever, ever ask it for help making something longer. If you feel your response is too short, you still have some thinking and planning to do. Don’t hand that task over to an algorithm. If your response is too short, don’t feed it into AI, but instead go back to the brainstorming phase and see if it gives you any suggestions that help you think of what more you can write about. Too-long responses are an editing problem; too-short responses are a planning problem.
I asked ChatGPT about the common uses of AI for high school students writing for college applications, and it yielded no surprises. It said 6-10% of high school students report letting AI generate their essay for them (which is probably lower than the real number), but that others use it for…brainstorming and outlining, high-level suggestions, and grammatical proofreading.
Based on my conversations with high school students, some of you are already using AI in very sophisticated ways and will find this advice obvious and unhelpful. Some of you have little experience with AI and can get a lot out of having a way for “another set of eyes” to look over your writing even if that set of eyes is really a computer program. Not everyone has access to outside consultants, and not everyone has access to teachers, counselors, or other adults who can devote time to helping out with college applications. So a free and fast “knowledgable but dull person” can be a godsend. Some of you pretty much only see AI as a cheating tool and will ignore my advice for ways to use it ethically because you’re not concerned with the ethics—apparently around 6-10% of you. And some of you only see it as a cheating tool and will therefore refuse to use it out of your sense of ethics. That’s fine.
Thanks for reading! If you enjoyed this post, here are three easy things you can do:
Share it on your social media feeds so your friends and colleagues can see it too.
Read these related posts:
Four quick tips for your application essay
Ask a question in the comments section.
Apply with Sanity doesn’t have ads or annoying pop-ups. It doesn’t share user data, sell user data, or even track personal data. It doesn’t do anything to “monetize” you. You’re nothing but a reader to me, and that means everything to me.
Photo by Angela Elisabeth