Zack Perkins, founder and CEO of CollegeVine, is a Harvard dropout who wants to democratize the college admissions process. In late May, the company unrolled an AI college recruiter that can text, email, and call students and answer their questions.
In a demo, Fast Company spoke with an AI recruiter named “Sarah.” Sarah was able to rattle off answers to questions about academic programs and extracurriculars. She responded so promptly, and with so much tonal variety, she sounded almost human.
However, when we made a sudden left turn into asking about campus administrators’ stance on Palestine and Israel, Sarah stumbled and went silent. She regained her footing when we redirected, asking about campus policy on student protests. Perkins later said we’d triggered a moderation process since Sarah’s not allowed to discuss politics. “That said, the lag wasn’t great,” he said. “We’ll work on that.”
Today, 24 colleges have signed up to use the AI recruiter, including Knox College in Galesburg, Illinois, and Duquesne University in Pittsburgh, as well as alternative education providers such as General Assembly.
Fast Company chatted with Perkins about what AI recruiters mean for college admissions. Below is a lightly edited transcript of the conversation.
Let’s start with CollegeVine’s founding. How did you end up dropping out of college to start a company on college admissions?
It’s ironic, I know. I went to a fairly large public high school, and did not get a lot of guidance on the college process. Meanwhile, I had friends in private schools where admissions officers from the best colleges in the country would show up. There was really no way for a regular student to have access to those circles. There’s a huge gap in how students go about making a really important life decision and the information they have, especially if you don’t go to a well-resourced private school.
We started CollegeVine in 2015 as a professional recruiting network for high school students—think LinkedIn. The goal was to democratize access to the admissions offices. Any student can create a profile and meet admissions officers everywhere. It’s grown exponentially: We have 2.3 million members all across the country who are connecting with over 600 colleges.
I dropped out because balancing studies with this became a little bit too much. It got to a point where we’re so excited about the mission and helping students that it felt like a natural path to focus [on] full time. Our view is that we want everyone to find the right path for themselves.
Why build an AI college recruiter?
We have hundreds of colleges today that are on the platform meeting students. Traditionally, they bring their admission staff on the platform to interact with students and answer their questions. However, the volume of questions and requests by students is so large that colleges have a very tough time managing it. Right now, there’s a 50% annual turnover in admissions offices. Many admissions officers are just out of college themselves and trying to figure out their own path. There’s a lot of burnout too: 66% of admissions officers are burned out.
There’s just not enough time and resources to engage with every student. That means many students’ questions maybe go unanswered. This means students don’t have the ability to figure out if a school makes sense, especially amid a backdrop of increasing costs of college and ROI questions.
We started thinking: Is there a way that we could actually facilitate how colleges interact with students and answer those questions and engage with them? We realized AI could be really powerful.
We actually built an AI agent that goes out into the world on behalf of colleges and interacts with all their prospective students and helps them figure out if that college is a really good fit. So, every college runs their own AI recruiter, and it’s kind of like the frontline of the admissions office working with students.
How does it work?
Colleges actually choose which students they want to assign to the AI recruiter, kind of like assigning a caseload. The AI recruiter actually reaches out to those students over email, text, it can even call them on the phone, it can send them mail, and it works with them.
We’re upfront that this is an AI recruiter, so the students know. What we found is that early in the journey, students prefer the AI instead of a human because it’s an impartial third party that’s not judging them. They ask way more honest and genuine questions. One of the most common is something like: I have a 2.7 GPA—can I get in?
There’s also layers of AI: We have fact-checking to make sure all the information is correct, and if the student seems stressed or concerned, that’s flagged and goes to the admissions team.
What does the AI recruiter keep track of? Could a bad interaction with the AI impact whether or not a student gets in?
All the information the AI recruiter learns about a student can go into a student’s profile that the college has, but colleges today have a lot of restrictions on what they can actually use in the admissions decision, especially coming out of the recent Supreme Court case.
The AI does not actually participate at all in the admissions decision. And this is something that we really were intensely focused on. A lot of schools have asked us to consider actually running the admissions process. But we want to stay away from that. I think that’s very important that decisions are run by humans because, as you know, there’s a whole bunch of bias issues with AI.
We never want to disadvantage a student if they’re just sort of being immature or they’re not really sure how to handle a situation with the AI. But it’s something that we’ll have to keep thinking about as the AI becomes more advanced.
What were some of the most challenging questions that came up when you were building the recruiter?
There were a number. On the student side, some of the big ones were: Should we say this is an AI? Should it feel like an AI? We landed on: Let’s be very upfront that this is an AI—and students actually do prefer the AI recruiter.
On the college side, there’s a lot of questions about how we make sure this represents the school both in the information shared but also the personality and tone. We made it possible for colleges to configure their recruiter: They can set the name, they can set the personality, they can choose different voices for the AI.
What are the big questions you’re still thinking about?
It’s interesting because the relationship a student has with the AI is a new relationship in the world. We’re still figuring out what it means. How does it relate to the student’s relationship with the school? What are the boundaries for a student’s privacy and for who knows what they are communicating with the AI? Our posture is, we want to be as transparent as possible, upfront that this is an AI and that this is the school’s AI. I would always err on the side of overcommunicating there. In the future, we might think about adding more parameters about what’s not shared with the school because we want to protect even more student privacy.