Steve Gardner has a direct answer for the parents who call him each summer. The college admissions consultant and Harvard Summer School instructor tells them their high-achieving teenagers probably should use artificial intelligence. But only in one specific way.
Gardner coaches students headed to Ivy League schools. He watches them mine their own lives for raw material before they touch a keyboard. Failures. Sacrifices. Contradictions. Those elements turn a forgettable essay into one that makes an exhausted admissions officer sit up. Only after a student produces a full, honest first draft does Gardner suggest turning to AI. Not as a writer. As a tough critic.
“AI won’t ever have your child’s insight, memory, or voice,” Gardner wrote in Business Insider. “What it could do, after a student has built an honest draft entirely on their own, is push that draft to be sharper.” Prompts such as “Does this part sound generic, or am I making assumptions I have not examined yet?” or “Is anything here confusing or underdeveloped?” turn the tool into a sparring partner. The thinking stays human.
This measured approach separates the students Gardner sends to elite colleges from those who treat ChatGPT like an essay mill. The distinction matters more than ever. A study of more than 370,000 applications found that after ChatGPT’s arrival, essays grew more sophisticated and colorful. They also began to sound alike. They read better and said less.
But the gap between smart users and careless ones runs deeper. Research from Cornell scholars published this spring shows lower-income students rely on AI more often. Those who use it face higher rejection rates. Higher-income applicants pair free or paid AI with counselors, teachers, and paid coaches. The result? Their polished output still carries personal texture. The digital divide didn’t disappear. It changed shape.
“High-income students have a lot of different resources; they have counselors, they have teachers, they have more support on top of ChatGPT,” Jinsook Lee, one of the Cornell researchers, told Inside Higher Ed. Lower-income students often work with basic versions and lack the human guidance that prevents homogenization. Essays from AI-heavy applicants lost the quirky, personal, even “idiotic” narratives that once revealed character.
Admissions offices know this shift is underway. Policies remain a patchwork. Among the top 30 universities, 21 have no official rule on AI. Only Brown and Georgetown ban it outright. Eight others, including Caltech, Cornell, Vanderbilt, Carnegie Mellon, WashU, Emory, and UVA, permit limited use for brainstorming or editing but forbid generating core content. The rest lean on honor codes and the Common App’s fraud policy. Applicants must attest that their work is their own. GradPilot reports that no top-10 school publicly uses AI detectors. Penn stands alone in stating it avoids AI entirely when evaluating applications.
Some institutions experiment on the other side of the desk. Virginia Tech now runs every essay through AI scoring alongside a human reader. Disagreements trigger a second human review. Other schools feed applications into AI to generate summary profiles for committees. The summaries flatten generic essays into invisibility. An essay that reads like every other AI-assisted piece disappears in the stack.
This creates a new pressure. Students who once paid consultants thousands of dollars for bespoke essays can now generate dozens of variations in minutes. The barrier to entry dropped. So did the signal of genuine effort. Admissions officers, reading 100 files a day in peak season, spot the patterns fast. Repetitive sentence rhythms. Overused reflective phrases. Vocabulary that screams large language model even if detectors stay silent.
Yet the best applicants treat AI as one instrument among many. They spend a week cataloging personal contradictions before drafting. They produce 10 or 15 full revisions. The AI critique arrives late, sharpening language without supplying the core ideas. That sequence preserves voice. It also mirrors how professionals in many fields now work. Lawyers draft briefs, then query AI for gaps. Writers produce manuscripts, then use tools to test clarity.
Gardner’s students who reach Ivy League schools follow this exact rhythm. They avoid lists of accomplishments. Those belong in the activities section. Instead they deliver insight an officer hasn’t encountered, reflection that feels earned, and a personality that sparks the thought: I would enjoy having this person in my dorm.
The broader market adapted quickly. New AI tools aimed at college applicants flooded the space this year. Some focus on brainstorming. Others claim to mimic successful essays. A few promise to optimize entire applications. Most admissions experts warn against full reliance. The resulting prose lacks the messiness of real teenage experience. It reads competent but forgettable.
Colleges themselves turned to AI for efficiency. Some use it to scan for keywords or flag incomplete files. A few, like certain University of California campuses, have signaled discomfort with sterile AI output. One UC Davis admissions leader told applicants that an AI-written personal insight question reveals nothing about the student’s hopes, trials, or context.
The arms race shows no sign of slowing. As detection improves and policies tighten, students who master strategic AI use may hold an edge. Those who over-rely on it risk rejection not just for dishonesty but for blandness. The memorable essay still wins. AI can help refine it. It cannot supply the lived material that makes it memorable.
Parents who call Gardner often fear their child will fall behind without AI. He flips the concern. The students who stand out treat the technology as a finisher, not a creator. They do the hard work of introspection first. They revise relentlessly. And when they finally ask the machine for feedback, they already know what makes their story theirs.
That approach demands more discipline than simply feeding a prompt and accepting the output. It also produces better results. In a season when thousands of applications arrive with similar SAT scores and GPAs, the essay that reveals a distinctive mind still carries weight. AI cannot invent that mind. It can only polish the words that describe it.
Recent surveys suggest half of applicants now use AI at some stage. The share generating first drafts hovers around 20 percent. Those numbers will climb. The question is whether students learn to wield the tool or let it wield them. The evidence from this year’s cycle suggests the former group keeps a clear advantage.
Admissions deans face their own balancing act. They want authentic voices. They also need to process record volumes without burning out staff. Some added essay prompts about AI ethics. Others quietly updated training for readers to recognize homogenized prose. A few dropped numeric essay scoring altogether, wary that AI-influenced rubrics would compound the problem.
The result is a more human process in some ways, more mechanical in others. Students who understand both sides gain the upper hand. They craft essays that pass human scrutiny and algorithmic filters. They reveal personality that no language model can replicate. And they do so by keeping the core work where it belongs: in their own hands.
WebProNews is an iEntry Publication