IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

University Admissions Use AI to Gauge Applicant Skills, Traits

A new artificial intelligence tool from Student Select can rate non-cognitive traits, such as positive attitude and conscientiousness, as well as performance skills like communication and critical thinking.

A laptop open on a desk with the word "Admissions" on the screen below a symbol of a school and above a box that says "click" and a line below it that reads "click here to study." Black text on a yellow background. There are numerous pens and pencils and books on the desk around the laptop, and there are hands on the laptop keyboard.
For graduate school candidates at Rutgers and Rocky Mountain universities, the application process could go much deeper than transcripts and standardized test scores. And yet, admissions officers at both schools expect to notify all applicants of their acceptance status sooner than competing institutions.

The two universities are the first to use AI software from the data science company Student Select in their admissions process. Will Rose, the company’s chief technology officer, said both schools signed on for the subscription-based service for the current academic year. Five other U.S. schools, whose names he could not disclose, are utilizing Student Select right now for fall 2023 applicant pools.

In an interview with Government Technology on Tuesday, Rose said the product in its current form is mainly useful to competitive graduate schools that want to take a closer look at “borderline” applicants who may have slightly lower grade point averages or test scores than the automatically admitted students, but enough to offer by way of leadership, communication and critical thinking to nudge them over the line.

“It’s not intended to make admissions decisions [for the school] at all,” Rose said, adding that the technology broadens yet streamlines the selection process, but it does not replace human admissions counselors. “The goal is to get the acceptance letters out faster.”

Rose said his company custom builds a machine learning model for customers based on their rubrics for selection, which can include grades, scores and any quantitative or qualitative data considered in the application process. The schools can determine if they want to omit test scores, such as the GRE or GMA, from the review process or lessen the weight of those scores when looking at overall qualifications of applicants.

He said admissions counselors use Student Select to establish three tiers of applicants: The first tier is those who exceed all criteria and are automatically admitted, the second pertains to borderline applications which trigger additional steps, and the third is made up of those who don’t meet minimum requirements and would be rejected.

For those tier two applicants, Student Select uses AI to review and score their essays and previously recorded interviews with admissions counselors. In a demonstration of the product, Rose noted that the product can rate non-cognitive traits, such as positive attitude and conscientiousness, as well as performance skills like communication, leadership and critical thinking. Based on the scores of those attributes, admissions officers can decide to accept or reject the candidate, conduct a second interview, or wait-list them for future consideration.

All applicants are notified of their status in a matter of days or weeks, not months.

Rose said Student Select reduces or even eliminates age, racial and socioeconomic bias in the consideration process because the technology does not look at dates, candidate names or ZIP codes. It can also analyze the school’s historic admissions data, which admissions officers or administrators can review to identify underrepresented groups. Rose maintained that, historically, higher test scores “correlate with affluent people,” and under that assumption, institutions can elect to replace standardized test scores in their rubrics with other qualifying data identified by Student Select.

“It’s not our recommendation whether schools replace [standardized test scores] with different data,” Rose said, “but programs want to do that.”

Rose said Student Select can also be built for the undergraduate selection process and interfaces with Common App, an online tool for high school students to access college applications. He added that Student Select has the potential to integrate with tools that verify the authenticity of submitted essays and identify improper use of ChatGPT.

Student Select has been on the market for about a year now, Rose said. He acknowledged that there are many software tools in the college admissions space, including a program from Kira Talent that is used only for analyzing applicant interviews, but he is unaware of others that use AI to identify and score an applicant’s insights and skills.

According to a video on its website, the value proposition of Student Select is its ability to “synthesize large amounts of information from diverse sources” while freeing up employees in short-staffed admissions offices to spend more face time with the prospective students. It also has the potential to scour an applicant’s social media presence and incorporate “facial analysis” of expressions and eye or lip movements from video interviews into the algorithm, but both measurements are considered controversial in the admissions process, so they are not currently in use, according to the video.

Editor's note: A previous version of this story said Will Rose could not disclose the names of five schools using Student Select because their contracts haven't been finalized, but it has been updated to explain that it's because their contracts prohibit disclosure.
Aaron Gifford has several years of professional writing experience, primarily with daily newspapers and specialty publications in upstate New York. He attended the University at Buffalo and is based in Cazenovia, NY.