Did AI kill the college admission essay?

If the latest artificial intelligence tools have brought powerful new capabilities to students and educators, they’ve also introduced confusion and raised new questions about academic integrity.
Recent survey results published by the admissions assessment and analytics provider Acuity Insights show that many college applicants are unclear on when they’re allowed to use AI. Thirty-nine percent of college applicants said they didn’t use AI as part of the application process because they believed it might be considered cheating.
Kelly Dore, who co-founded the company, said that both students and university leaders are confused about AI’s role in higher education, and that she believes skills assessments like those her company sells can provide at least one workaround for AI’s chaotic influence.
“Almost 65% were confused about the extent to which they could use AI in their admissions and what we see is that increasingly there’s an imbalance of the use of AI,” said Dore, an adjunct associate professor at McMaster University at Ontario, Canada.
Dore said there’s a rift between students who don’t use AI at all and those who use it heavily or moderately. She said schools, for their part, could help clarify things by being more transparent about what students and applicants are allowed to do with AI, as well as disclose how their own institutions are using AI.
The University of North Carolina at Chapel Hill was recently reported to be among the institutions using AI to sort through applications, automatically scoring the writing quality of essays. Texas A&M University and Georgia Tech are also reportedly among those using AI to streamline application evaluations, scanning for keywords and using machine learning algorithms to find the most promising students.
Dore’s business began with the creation of an assessment she originally co-developed for medical schools, designed to test the soft skills of pre-med students, skills like professionalism, ethics, communication and empathy. Since developing the test — called the Computer-Based Assessment for Sampling Personal Characteristics, or Casper — Dore has seen it adopted not only by admissions offices of dozens of medical schools but by many other types of schools in the U.S., Canada, U.K., Australia and New Zealand.
Dore began work on the Casper assessment in 2005, alongside McMaster Oncology Professor Harold Reiter, at the request of McMaster University’s medical program, which was struggling to glean the insights it wanted from personal statements as it sought applicants who were not only smart, but showed the potential to have good bedside manner.
“They were having the problem of they were being written by other people, they didn’t actually predict anything and they took an enormous amount of time to read,” Dore said.
Even before the recent wave of generative AI tools, she said, many institutions have struggled to accurately measure students’ soft skills through examination of personal statements and reference letters, in part because they’re often drafted by groups of people, like applicants’ family and friends, and don’t provide an accurate reflection of an applicant’s character.
In addition to focusing on real-time reactions, as opposed to allowing applicants weeks to craft the ideal persona, Casper also markets itself as a standardized measurement designed to reduce some of the biases found in assessment of letters, such as a tendency to favor privileged students with larger social networks who have greater access to prominent people who can recommend them.
One meta-analysis published in 2014 by the International Journal of Selection and Assessment showed that letters of recommendation have a weakly positive correlation with performance in higher education, though direct connections to soft skills are less clear. An ongoing study aims to measure Casper’s ability to predict nursing student performance.
But whatever assessment tools schools use, AI isn’t going away, Dore pointed out, and many believe demand for soft skills will likewise grow as computers become more competent at a growing number of technical tasks.
“I think [soft skills] are only going to become more important with AI,” she said. “I think it fundamentally changes what we need to focus on when we’re educating students, what we’re preparing them for in the workforce going forward. All of those things are shifting with AI.”