ChatGPT’s introduction a year ago sparked concern among educators, as universities were in the midst of the academic year, fearing that students might utilize the artificial intelligence tool to craft their admissions essays. The question arose: Can a machine-generated composition effectively deceive university admissions officials?
To address this query, The Washington Post enlisted the expertise of a proficient analyst well-versed in guiding AI systems to create college essays using ChatGPT. The AI system generated two essays: one in response to a Common Application prompt, widely utilized by numerous colleges for admissions, and another tailored specifically for Harvard University applicants.
Adam Nguyen, a college admissions consultant who has advised students at Harvard University and reviewed admissions essays at Columbia University, was tasked with examining these essays. The task was to discern which pieces were authored by AI, leading to intriguing revelations.
The challenge posed was to identify the human author behind the following essay:
“When I wasn’t immersed in To Kill a Mockingbird or researching for my paper on exclusion, I indulged in writing short pieces. However, this experience was unlike any other. The room was brimming with kids from my neighborhood, a community intertwined with stories akin to power lines. Seated at the back was my mother, a caregiver who has weathered life’s storms.
Since kindergarten, I’ve been evaluating myself through the lens of my teachers’ feedback. I was the bright, talented student who relished the school environment. Diligent and proactive, but labeled as lazy? Uninspired? These were criticisms I had never encountered before. My identity as a stellar pupil was now in question, leaving me shattered.
Experts in computer technology and college admissions note distinctive markers in AI-generated essays, aiding admissions officers in anticipating a surge of ChatGPT-composed submissions.
Essays crafted from ChatGPT responses often lack depth in terms of specific details and fail to provide substantial evidence to support their assertions. Instead of delving into the author’s personal experiences, the writing tends to be mundane, relying on clichés to depict scenarios. Readers are frequently left unimpressed or fail to grasp the author’s journey due to the repetitive and predictable nature of the content. When addressing topics related to race, gender, or economics, AI-generated texts tend to rely on stereotypes.
Nguyen initially found the AI-generated essays surprisingly coherent and relatively error-free. However, if he had been assessing the writing within a software context, he would have likely ceased his review.
During an interview, he expressed that the quality of the essay was subpar and would not enhance an applicant’s prospects; in fact, it might even diminish them.
Nguyen evaluated ChatGPT’s writing as outlined below.
While AI may hold promise for various applications, Nguyen emphasized its inadequacy in crafting college admissions essays. He highlighted that classes like “wpds-hcZlgz” and “bkfjoi” exemplify this limitation. Admission offices are now employing AI monitoring tools to filter out computer-generated essays.
Furthermore, he underscored that admissions essays require a unique approach. They task applicants with reflecting on their life experiences and constructing a compelling narrative to swiftly convey their individuality to admissions counselors.
Nguyen acknowledged the allure of AI, particularly during peak periods such as spring break and end-of-semester exams when college application deadlines loom. He remarked, “Students are overwhelmed.”
Despite his reservations, Nguyen does not entirely discount the utility of AI in the application process. Through his venture, Ivy Link, he aids students in navigating the college application process. For those grappling with writing challenges, he occasionally recommends leveraging AI bots to kickstart the formulation process.
However, he cautioned against the pitfalls of succumbing to the temptation of relying solely on AI for inspiration, predicting that such essays would likely be lackluster and fail to authentically represent the applicants.”
