Near the start of the summer, Zack Perkins and his technology company, CollegeVine, released a product-launch video that sought to show the promise of artificial intelligence in scaling up the work of admissions offices that give information to prospective students.
Perkins, dressed in a black T-shirt and jeans, announced “a new era of recruitment” by taking a call from “Sarah,” and asking her to introduce herself.
“I’m Sarah, an AI recruiter from Monza University,” the bot said in a vaguely seductive voice, representing a fictional college, created for demonstration purposes. “I meet prospective students, have amazing conversations about their interests and goals, and get them really excited about what we have to offer.” Sarah asked Perkins, posing as a student: What was most important to him in his college decision?
“I think I really want to be an accounting major, and I also want to make sure I get a good job after I graduate,” Perkins said.
“That’s great to hear, Zack! Monza University has an excellent accounting program that prepares students for successful careers. Our career-services team works closely with students to help them secure internships and job opportunities.”
I had to try it. I registered on CollegeVine’s site, and within moments Sarah from Monza called my cell phone. When she asked what I was interested in pursuing, I said anime, and Sarah guided me to Monza’s animation program. When I said I wasn’t really interested in the art of anime but rather the stories, Sarah shifted gears and guided me to a program in Asian literature. Did that sound more interesting?
It was an impressive feat of technology, but it raised more questions: As part of a demo program, “Monza University” has limitless academic programs — so what would Sarah say to a student applying to a college that doesn’t have animation or Asian literature? And for the student interested in accounting, would an AI assistant ever ask, “Why accounting? Do you really know what accounting is?” And would it pick up on the subtleties or unspoken elements in the student’s response?
Those are crucial questions, because in all likelihood, AI assistants will be everywhere in a matter of years — in recruiting, career coaching, academic advising, tutoring, and more. In addition to its AI recruiter, CollegeVine also has an AI-powered chatbot called Ivy that answers students’ general questions about what colleges they might consider applying to, what major they should choose, and what they might do with that major. (CollegeVine recently changed the chatbot’s name to Sage because, the company says, the new name “evokes a sense of wisdom, knowledge, and good judgment.”)
Having just written a book about the fundamental value of relationships, deep questions, and conversations in planning successful undergraduate degrees, I’ve become attuned to what AI might mean for the jobs of staff members and the work they do as a campus’s connective tissue. Many of them want to work more closely with students, rather than just push them to pick a major, check off other requirements on the degree audit, and process them through the system to graduation. In that, AI holds promise: Just as the ATM transformed the job of bank employee to less of a money dispenser and more of a customer-relations role, AI could handle rote work to free advising and coaching staff to get past the transactional interactions and on to conversations that unpack a student’s interests and potential — and craft a college plan.
Or you could imagine a bleaker future: In offering ready-made college shortlists, would AI shut down the exploration of some students? Would it reinforce conventional thinking about what majors can lead to? Most of all, would some college leaders — pressured to reduce costs, particularly at institutions that serve the neediest students — simply see these interactive tools as a way to replace personnel with a machine?
AI offers all kinds of efficiencies to the rising workload of admissions. Nathan Ament, vice president for enrollment at Knox College, had already been a customer of CollegeVine’s student-networking platform, so he got a preview of its AI admissions recruiter in February. “I saw this as really revolutionary just because it was external facing,” while other AI solutions mainly focused on internal processes, like transcript previews or essay reading, he says.
Knox will use the AI — customized for the college and named “KC” — at the top of the admissions funnel, helping to convert prospects into inquiries. “My end goal is to try to get students to interact with it on the phone because we’re not doing anything with phone calls,” says Ament. Knox abandoned cold-calling years ago; the call-center companies were too expensive and admissions counselors spend more time managing the growing pile of applications than conducting outreach. “So I thought, What do we have to lose?”
Each institution’s AI recruiter learns about the college by “reading” its public-facing documents and “talking” with the staff. The college spent the last part of June and all of July scraping information from the website to give to KC, and then allowing “her” to have one-on-one phone conversations with all the members of the admissions staff, feeding yet more information to the machine. From 35 choices, institutions can pick the name, gender, and voice qualities of its AI recruiter, and KC is female — a decision Ament says wasn’t necessarily intentional. “It was already female, and everybody was used to interacting with the female voice on the other end, so we just went with it.”
The school is still in control, in that they own the model, they own the training. It’s their employee, basically.
Perkins, of CollegeVine, said in an interview that of the 77 institutions that have signed up for the AI recruiters (half of which are now in use), many started out using a formal voice but shifted to something less formal — more like a peer — over time. After a student interacts with the AI recruiter, the institution can set a series of follow-up connections via text or email — Perkins calls them “affinity events” — to try to hook them in. Many institutions had “generic” follow-up plans, says Perkins, asking the student to read more about the college or learn about some aspect of its offerings. Perkins says his company has observed that students are focused on “a checklist of things they’re trying to qualify in their head before they decide it’s a good fit”: What does it take to get in, what does it cost, and does it have my desired area of study?
“So we’re going to reconfigure your affinity events to help check those boxes, which is how you actually get that buy-in,” Perkins says. (Often, he says, students are more willing to ask the bot about things like GPA requirements, because it’s not going to judge them.)
After the AI recruiter trains on the materials, it runs through thousands of simulated student interactions to evaluate its attention to accuracy and relevance, and it continues to train after it goes live. “If it ever doesn’t know the answer to something, it will say, That’s a great question and I actually don’t know that answer. I’m going to check with my team and I’ll get back to you” — and that will trigger a cue to schedule training with college personnel.
“But the school is still in control, in that they own the model, they own the training,” Perkins says. “It’s their employee, basically.”
As college costs have become more of a concern, higher education’s critics (and often its faculty) have found a plump target in the expansion of management and staff roles — or “administrative bloat.” In a recent edition of his newsletter, Ryan Craig argued that AI would “shrink the university” — but not necessarily in teaching and learning, which is getting most of the attention. “The money is spent on everything else,” wrote Craig, a managing director at the education-focused venture firm Achieve Partners and a frequent critic of the inefficiencies of higher education.
Citing Gordon Gekko on corporate vice presidents who shuffle paperwork, Craig predicts that AI “will end busywork” in higher ed.
“There’s much more busywork to end outside the classroom,” he wrote, listing two dozen areas that stand to gain from AI’s ability to automate routine tasks, see patterns, anticipate problems, and improve service: admissions, career services, facilities, human resources, residential life, student judicial affairs, and student success and retention, among others. Craig is a skeptic of the work that some of these support services do — for example, he coined the term “career-services theater” to describe what he sees as high-profile but ineffective career coaching on campuses.
Many people working in admissions worry about how AI will affect their jobs. But a bigger fear, says Emily Pacheco, an assistant director of undergraduate admission at Loyola University Chicago, who was one of the founders of a growing LinkedIn group focused on AI in college admissions, is that people in her profession will ignore or resist grappling with the technology, missing the ways to make the best use of it.
A lot of our work in admissions is so repetitive, and we’re getting burned out.
The busywork is undeniable, she says. Admissions often becomes a catch-all for student questions, because its email address and office phone number are usually listed prominently on college websites. Each day, the admissions team at Loyola University Chicago assigns someone to spend five to seven hours answering questions about major offerings, the Catholic organizations on campus, or the crime stats around the neighborhood. Some of the questions are completely misdirected — a student, for example, asking how to register for a required course that has filled up.
“I’ve got two college degrees, and I find myself going to Google and asking what percentage of students are living on campus, because I can’t remember,” says Pacheco.
“So much of my work time is spent doing that kind of work rather than that higher-level work, that real personal connection,” she adds. “A lot of our work in admissions is so repetitive, and we’re getting burned out.”
Even in personal interactions, Pacheco says, “there are a large percentage of my conversations where I feel like a glorified Google.”
But the dividing line between busywork and meaningful interaction isn’t always clear. She recounts a recent 45-minute conversation with a student, the first 30 of which “probably could have been done by a bot.” But in the last 15, the student started talking about her emotional struggles, the problems she had in high school, and how the university might address her learning disabilities.
Could that student have interacted with a chatbot first, then gone to Pacheco for a more-human connection after? Or did the first 30 minutes with Pacheco help that student get comfortable enough to open up? Good counselors and advisers have all sorts of techniques for making these moments happen, by asking unexpected questions, or using humor, or code-switching — things AI doesn’t do well, at least for now. Identifying those moments, those triggers, will be crucial for college leaders designing services around these tools.
An open question centers on whether these AI tools can be as resourceful in helping students as professionals in the field, who often make these connections intuitively. A good counselor can connect needy students to hidden assets or loopholes that help them pay for college or gain an advantage — how a student who was on a free-lunch program in high school, for example, can be exempt from mandatory fees, helping to make matriculation an easier decision. Helping a student hack the system in this way sometimes depends on the institution better understanding a life story or quality that it might not track closely, and that AI is ill-equipped to draw out.
Sebastian Brown, a co-founder of the admissions LinkedIn group, says that he sees ways that AI tools could be used to analyze the hundreds of variables among applicants, helping an institution see patterns in the noise.
“Folks are trying to use AI to better understand how some of these variables might be interconnected in ways that we didn’t realize before,” says Brown, who works as a regional admissions counselor for the University of Oregon, but said he was speaking only in his capacity as a manager of the LinkedIn group. “Just that additional power from AI to be able to dig a little deeper into the data is where that can be powerful.” But effectively executing on the data — and keeping a watch on the bias an algorithm might introduce — could still require human interaction.
Brown grew up in an affluent suburb of Denver and attended the University of Redlands, an institution that was off his radar but proved to be an ideal fit, where he earned both a bachelor’s and a master’s degree. He wonders if AI will be good at finding more ideal fits for students, or whether it will reinforce set patterns. “Just playing around with AI and building those college lists, that’s a school that’s almost never going to end up on someone’s list,” says Brown. For now, he believes students more typically find the best institutions through a knowledgeable counselor, an education consultant, or a random encounter at a college fair.
It’s possible, however, that human high-school counselors or mentors could use AI to expand their knowledge of what’s out there — feeding a list of colleges to the computer and seeing what other related, relevant colleges it can pick out. Pacheco spoke with an independent education consultant, or IEC, at a recent conference who had been in the field only two years and didn’t feel she had yet gathered enough knowledge to find “good fits” for her clients.
Using an AI tool, Pacheco advised the consultant, “turns you from a two-year IEC into an IEC with 10 years of experience.”
Picking a college is stressful enough. Once in college, a student faces an array of choices about majors, minors, and electives that have implications for a life trajectory. Because of the cost of college, many students feel pressured to pick a “secure” major and pick it early.
“It’s an impressionable time in people’s lives, obviously,” says Alex Sevilla, the vice provost for career advancement and engagement and executive director of the career center at Vanderbilt University. Before working in career services, Sevilla ran an MBA program at the University of Florida, where he saw many students enrolling to retool their lives after making wrong turns as undergraduates. They had chosen a path that they hadn’t thought through and found themselves in a career track that wasn’t what they wanted.
“Would AI make that less of a proposition or more of a proposition? That’s an open-ended question,” he says, noting that students already come in with limited perspectives on what careers are possible — which affects how they might question the bot, yielding answers that might obscure the possibilities of a less-trod route to a good career and meaningful life. “Would we create more college graduates that would feel a hollow level of comfort in a pathway?”
We’ve gotten to the point where we can tell who has put a résumé and a cover letter through ChatGPT, because it has zero humanization in it.
On the other side of the college-to-career arc, all sorts of AI tools will help students write cover letters and résumés to head out into the world, but career centers are finding that students still need help to brainstorm ideas, add creativity, and give their application materials depth. “We’ve gotten to the point where we can tell who has put a résumé and a cover letter through ChatGPT, because it has zero humanization in it,” says Joseph Catrino, executive director of career and life design at Connecticut’s Trinity College, who has been writing about AI recently. “Actually, we’re talking to employers who are now able to tell when students are putting it through — and that’s problematic.”
CollegeVine’s chatbot, Ivy, helps students find institutions, pick majors, and investigate career paths. When asked about institutions in the Midwest with good environmental programs, Ivy offered up Gustavus Adolphus and Northland Colleges — two small colleges I thought it might overlook.
It gave good but very general starting suggestions for how to pursue a career in environmental art and child psychology, offered decent advice about the basics of networking and the job search, and outlined solid (but, again, generic) plans for transferring from a community college to a four-year institution.
But in helping to pick a major, Ivy made some questionable choices: When I told Ivy I wanted to start a small business and needed advice on a major, it offered up business administration or management, and suggested looking into experiential learning and specific tracks in business (marketing, finance, entrepreneurship) that might line up with my goals. When I added that the small business I wanted to start was a martial-arts studio, it recommended largely the same path in business, but to keep up my martial-arts training on the side. When I changed the prompt to say that I might rather open a running-shoe store, Ivy again offered up a business major but suggested attending trade shows and brushing up on the shoe industry.
But someone who wants to open a dojo or a specialty shoe shop could major in kinesiology, or biomechanics, or a health field — it depends on one’s angle. Business is just the most conventional major-equals-job route. More granular training in one of those areas might better fit a student’s abilities and talents — or true interests. Training in kinesiology or biomechanics might allow a student to enter the world of martial arts or running in a way that a business major cannot.
Perkins says Ivy is built on data about college and the information CollegeVine pulls from the student profiles on its site, which tracks students from enrollment through college to a job. That information is put through a machine-learning model, then contextualized for human users through the counseling strategies the bot has learned.
“It’s observing — I suppose in an aggregate, anonymized way — what is happening across all these different students,” he said, and then using those patterns to make predictions for other students.
Perkins readily acknowledged that in learning from the trajectories of CollegeVine profiles, Ivy might not pick up on the idiosyncrasies, hidden techniques, and unusual patterns that nevertheless led to success — the student, for example, who gets a GED but goes on to earn high scores at a community college, and uses a compelling personal story and work experience to gain admission to an elite university.
That hacking of the system often happens between humans. In the best instances, advising and mentorship is a creative act: It’s not that the adviser or mentor creates the path for the student (which does happen, for expediency), but that the mentor draws out the student’s hidden talents and goals, teaches them how to see the learning possibilities everywhere, and helps the student create the path that fits their strengths. Asking advice from AI can be one technique in this process — but students would have to be savvy enough to know what queries would reveal those hidden paths and yet-unseen possibilities.
Many students already get inadequate, transactional advice from admissions, academic advising, and career services — not just because of the busywork, but also because even some people who work for colleges have limited ideas of what a college education is about. Perkins says that many of the institutions that CollegeVine worked with “had a very low-resolution view of their offerings, and so they were not providing a lot of information.” His company had to build the AI recruiter with a desire to learn more about the institution.
Either way, the most optimistic future of AI is likely to involve liberating the work of people and what they can do best: synthesis, connection, and summoning the hidden strengths and commitments they find in students to help them take advantage of the higher-education system. College staff who don’t embrace those roles might find themselves like auto workers, retail cashiers, and others whose jobs were destroyed by automation.
“Our mission statement is to move away from transaction and move toward transformational, but we really want these transformational conversations to happen with our coaches,” says Catrino, of Trinity. “Students come to us because it’s not just about career. We’re working with students through their executive-functioning skills, navigating problems. You can take that into a bot all you want, but you’re never going to get to the point where you’re going to have a human talking to you, really empathizing, and helping you understand.”
