This story about AI in college counseling was produced by The Hechinger Report, a nonprofit, independent news organization focused on inequality and innovation in education. Sign up for The Hechinger Report’s higher education newsletter.
By Jon Marcus, The Hechinger Report
At 16, Khloe Watson-Barrett already knows she wants to be a lawyer. She also knows she’ll soon have to run the gauntlet of the high-stakes college admissions process, now that she’s past the halfway point of her junior year in high school.
“It’s nerve-wracking,” Watson-Barrett said of what she’s heard about applying to college.
That unease is only aggravated for many high school students by a lack of access to one-on-one time with overburdened college counselors, who are often buried under questions about what tests to take, which deadlines to meet, and how to fill out applications for financial aid.
Now Watson-Barrett’s school and many others are just at the start of testing a new generation of technology that promises to free up time for college counselors while giving students crucial information, even outside school hours: artificial intelligence built specifically to provide advice about life after high school.
The AI being piloted for high school college counseling doesn’t simply skim the internet like general-purpose AI, which is subject to misinformation and manipulation. It’s programmed with answers provided by experts, based on a history of previous applicants’ questions. Unlike AI tools in many fields, several educators said, the AI being designed for college counseling — including a new AI platform called CounselorGPT that Watson-Barrett’s school expects to start testing next year — has the goal of encouraging more human interaction, not less.
College counselors still are best at understanding and suggesting how students can achieve their individual ambitions, said Diana Moldovan, director of college and career placement at the public Urban Assembly School for Law and Justice, where Watson-Barrett goes. “You can’t replace the trust,” Moldovan said.
But counselors get bogged down answering procedural questions or prodding students to complete tasks, she said. “If AI could do some of these things, that leaves more time” to talk with soon-to-be graduates about their academic, social, and financial choices.
Nearly half of students are already using AI on their own to navigate the complex process of applying to college, according to a survey released in February by the higher education consulting firm EAB. That includes helping them pick and compare schools, complete their applications, and prepare for standardized tests.
This alarms college counseling professionals. Most generative AI that families are using can’t detect misinformation or disinformation, they say, and is vulnerable to recruiters who use tricks to put certain universities and colleges at the top of search results.
“I would not want a young person to be using these tools by themselves, because it’s about asking the right questions,” said Angel Pérez, CEO of the National Association for College Admission Counseling, or NACAC.
“Ask [generative] AI what college you’re going to get into — you’re not going to get the right answers. There still has to be the human component,” he said.
Related: Interested in innovations in higher education? Subscribe to our free biweekly higher education newsletter.
Using AI to ease the burden on college counselor caseloads
Moldovan’s office, hung with the banners of colleges and universities, is often busy, though the school has only about 200 juniors and seniors, or fewer than 70 for each of the three college advisers — a much smaller-than-typical number.
Even these counselors are frustrated by having to spend time on things AI could do, such as helping students fill out financial aid forms and reminding them to write their application essays. And that’s with caseloads much smaller than the national ratio of students to counselors, according to the American School Counseling Association: 372:1.
Nationally, nearly 1 in 5 high schools have no college counselors at all.
College counselors in schools that do have them are responsible for so many other tasks that only about a fifth of their time is spent directly on college admissions advising, NACAC estimates. And they’re available only during school days and hours — not on weekends, holidays, or in the summers, when students’ questions might crop up.
These are longstanding problems purpose-built AI could help solve, said Pérez, who said AI tools should be able to provide objective information from admissions experts such as how to take the SAT and fill out the free application for federal student aid, or FAFSA .
“All of that can be given to students at the press of a button,” Pérez said, “so college counselors can talk about fit and match and the unique aspects of the individual student.”
Several different and competing versions of AI college counseling tools are in their early or experimental stages.
One is CounselorGPT, operated by Urban Assembly, a nonprofit that supports 22 schools in New York City including Watson-Barrett’s. So far the organization has tested the tool in 13 schools, an effort started before New York City unveiled its preliminary AI guidance for schools this week. City officials are expected by June to finalize a policy for deciding what kinds of AI platforms will and will not be allowed in schools.
Developed by and proprietary to Urban Assembly, CounselorGPT uses real-time job postings analyzed by the labor market data company Lightcast to show what jobs are in demand, how much they pay, what credentials they require, and how much it will cost to get those skills.
That’s information high school college counselors haven’t historically been able to provide but that consumers want, said David Adams, CEO of Urban Assembly, whose schools are expressly focused on preparing students for the workforce.
“Even when students have access to the highest-quality college counselors, they can’t possibly know all the information around labor markets,” said Adams. “Students are left guessing about what kind of degree leads to social and economic mobility.”
They often make the wrong decisions, he said. “They shouldn’t have to go through college and struggle to get a job before they realize the low return on that credential.”
Another AI platform, called the Expert Virtual Assistant, or EVA, is being developed by a private company, the College Guidance Network, with which NACAC has a partnership.
It uses AI as a starting point that, in response to questions, not only provides basic answers but takes users to videos and other resources about applying to college.
Jon Carson, CEO of the College Guidance Network, said he co-founded the company when his own son, then in high school in the affluent Boston suburb where they live, got only an hour a year with his college counselor, who had myriad other responsibilities, including dealing with students’ non-college-application-related problems.
“I was stunned,” said Carson. “That’s the quantity problem. Then we get to the quality problem, which is that the average counselor also has to be a social worker.”
AI can’t substitute for human college counselors, he said, but it can offer basic information, at any time. EVA also records what questions students ask it, so counselors can track them. Carson compares it to the time-saving form that patients fill out before they see a doctor.
“What you’re trying to leverage here is that one hour that the counselor does have,” Carson said. “And they don’t have to use it answering the simplest questions.”
Twenty high schools are testing EVA, he said.
Not all college counselors embrace this idea. Though many are willing to use AI to streamline their own duties, fewer than 40% see it as a way to provide services directly to students, according to a survey by the American School Counselor Association and researchers at Ball State University.
“It’s still in the very early stages, and I’m seeing a little hesitation,” said Pérez. He said some counselors worry that AI might replace them altogether, or be used as an excuse for raising student-to-counselor ratios even higher.
There are also fears that students can become attached to and overly reliant on AI. EVA originally was given a female icon and pronouns to go with its female name, but is being changed to an owl to avoid users sentimentalizing it.
Jaheem Shaw, a senior at the Urban Assembly School for Law and Justice, is deciding from among several universities that accepted him, with his counselors’ help. Like junior Khloe Watson-Barrett, he used the term “nerve-racking” to describe the process.
But he felt he benefited from the personal interaction with his college advisers and how they learned about him, his interests, and his shortlist of school choices.
Those advisers also gave him something else, he said: encouragement.
“And I don’t think that’s something you can get from AI,” Shaw said.
Contact writer Jon Marcus at 212-678-7556, jmarcus@hechingerreport.org or jpm.82 on Signal.
