Michelle Lohman isn’t a stranger to the transfer process.
For three years, starting in 2008, she seesawed back and forth between Northampton Community College, in Pennsylvania, and four-year institutions in the state. Each time, she felt untethered — and at the mercy of the institution she was hoping to attend.
“My status and credits felt like a mystery until getting the official answer,” she recalled. “I was uneducated about how my credits could transfer.”
The transfer process — especially the time-intensive and opaque task of determining whether, and how, credits earned at one institution may apply at another — is one of higher education’s most stubborn challenges. So for those like Lohman, who’s now assistant director of advising and transfer services at Northampton, making that task easier for administrators and students alike is a priority that requires innovative thinking.
And lately that’s meant looking to AI-supported tools for help. Tools that can, for example, analyze thousands of courses and suggest new matches. Platforms that can consume course information from a variety of sources and spit out a road map for degree completion. Or chatbots that close information gaps.
“AI is definitely a hot topic in higher ed and is definitely something transfer professionals should be thinking about,” said Emily Kittrell, assistant director at the National Institute for the Study of Transfer Students at the University of North Georgia. And “not just in terms of streamlining processes … but also what could be as the tech progresses.”
The intersection of technology and student transfer does seem to be of increasing interest: Sessions on AI use in areas like transfer “filled the most seats” at the recent annual meeting of the American Association of Collegiate Registrars and Admissions Officers, a spokesperson said.
Transfer students make up a sizable chunk of the higher-ed population: In the fall of 2023, 13.2 percent of all undergraduates were transfer students, according to the National Student Clearinghouse Research Center. But that share has the potential to be even larger. Of the nearly 80 percent of students who enter community college intending to transfer to a four-year institution, only about a third make the leap.
Factors that contribute to that disparity — “diffuse or inaccurate” information, as Kittrell put it, around credit transferability, and lack of timely and centralized resources for students to make informed decisions — aren’t particularly “sexy” challenges to solve, Isaiah Vance, assistant vice chancellor for advising for the Texas A&M University system, acknowledged. And the solutions require time and people that some colleges may not have to build the necessary infrastructure.
That is exactly why, sources say, these are the kinds of challenges AI may be well-suited to tackle.
The Articulation Challenge
Some of the most common uses of AI in the transfer process have been chatbots that answer recurrent questions around transfer in real time and direct students to relevant resources, Kittrell said.
The piece of the transfer process that sources say is particularly ripe for AI support, though, is credit articulation. At a high level, “articulation” is the process by which faculty members and/or administrators determine whether courses at another institution are equivalent to courses at one’s own institution, and decide to formally accept those outside course credits.
The reality is that the sector “was not designed for students to transfer between schools, so there are times it doesn’t make sense.”
That matchmaking process can be cumbersome. Course information is often fractured and inconsistent across colleges — especially in the many states where colleges aren’t required to have the same course-numbering system for lower-division and core courses. Existing articulation rules and agreements, too, may not be evident to students, hindering their ability to determine which colleges can shorten their time to a degree.
The reality is that the sector “was not designed for students to transfer between schools,” Vance said, “so there are times it doesn’t make sense.”
But what if AI could make that matchmaking easier?
An initial cohort of 57 two- and four-year colleges across the country are starting to experiment with this concept through the donor-funded AI Transfer and Articulation Infrastructure Network.
Cohort members receive free access to CourseWise, an AI-driven platform that is trained on the participating institutions’ course catalogs and existing articulation and course-equivalency data. The platform allows cohort members to both look up how one of their courses might transfer to another member institution and, conversely, how another member’s courses might align with courses in one’s own catalog.
The platform either identifies an established match, if it finds existing articulation rules and agreements in its data set, or suggests a new one. In the latter case, the underlying AI algorithm largely relies on course descriptions and natural-language processing — the use of complex math to identify relationships and similarities between words — to flag courses that might be equivalent to one another. (The hope is to add in even more information to inform suggestions, like course objectives and syllabi).
“This isn’t something where we’re looking at the number of keyword matches,” said Angikaar Singh Chana, chief operating officer at Equivalence Systems, which manages CourseWise. Rather, it’s a more-holistic analysis “using state-of-the-art techniques that look at really understanding, ‘What does this description mean?’”
For participants like East Stroudsburg University of Pennsylvania and DePaul University — both of which have articulation agreements mostly with in-state colleges — the newly minted network will hopefully not only expand the universe of students coming to their institutions but also help them provide their existing students with more options to stay on track.
Even if universities like DePaul “had all the power in the world to evaluate everything themselves, it would take forever,” said Nicholas DeFalco, its executive director of transfer recruitment and admissions. “So the chance to collaborate in a very meaningful way across the country is really transformational.” About a third of the private university’s incoming fall class every year is transfer students.
Extending the network feels especially important for Northampton Community College. As a public two-year college that prepares students to matriculate to four-year institutions, “we have to pay very close attention and be very mindful of the way our credits are accepted at other places,” Lohman said.
‘A Network Effect’
Another effort to close articulation gaps and expand the roster of transferable courses between colleges is happening at Arizona State University, which is testing an alternative, grant-funded approach: Its “Triangulator” tool. (About 20 institutions nationwide are piloting it, and there are plans to recruit more.)
Through the tool, AI algorithms analyze articulation rules and agreements, along with course catalogs, and make suggestions for course equivalents based not just on what is known about a given course, but what is known about how that course transfers across other institutions.
Visualize a triangle with a course from a different college at each vertex. If College A has an articulation rule with College B for a specific course, and also has an articulation rule for that same course with College C, then perhaps an articulation rule would make sense between the courses at College B and College C.
“It’s a network effect,” said Katherine Antonucci, director of special projects in the provost’s office at Arizona State. “If you keep adding more institutions, more rules, you’re going to be able to find more connections.”
Antonucci and her colleagues are already seeing the difference in productivity and time savings. In two hours of testing, an ASU team was able to review potential course equivalents for about seven times as many courses as the team would’ve been able to do manually.
These tools aren’t static; sources working on both CourseWise and the Triangulator intend to add more student-facing features down the line, for example. There’s also emphasis on improving transparency for students around the applicability of course credit, or more specifically, whether credits earned at one institution count toward general-education, elective, or major requirements at another based on a student’s chosen major.
“If a student has a ton of credit that’s transferable but it doesn’t apply” toward their degree requirements at their new institution, “you’ll still have to take this whole other chunk of classes,” said Vance, the assistant vice chancellor at Texas A&M system. No one wants, say, 60 credits getting transferred over only as electives.
Texas A&M pays for nine of its 11 universities to use Transfer Equivalency Self-Service, a tool from the software company Ellucian. The tool’s underlying AI algorithm takes inputted user information — like transcript details and anticipated major — and cross-references it against various training materials, including Texas’ core curriculum and common-numbering system database, the course catalogs of the system’s universities, and course equivalents from out-of-state colleges that are added to the tool’s knowledge base as they’re established. It then generates a report of how the student’s credits would apply at the Texas A&M institution they’ve selected and the courses they’d still need to take to complete their chosen degree.
The tool is public and accessible to administrators and students alike, Vance said. He added that a next step would be allowing students to upload transcripts as a PDF.
Preserving the Human Touch
Potential benefits aside, any use of AI in a high-risk setting is bound to raise questions, especially around how to maintain control and safeguard privacy.
Those using these tools said the fact that the training data is public (like course catalogs) and not currently tied to personally identifiable information helps alleviate concerns around student data privacy. They also underscored that suggestions for new articulations, in particular, are just that: suggestions. Establishing a new articulation rule or agreement still requires a human’s sign off, and users are encouraged to give real-time feedback when a tool gives a bad recommendation.
As Vance sees it, it’s not about removing humans’ involvement. It’s about redefining their roles — from being creators to being reviewers and editors. That shift, he believes, can preserve time and energy for circumstances that do require the human touch.
Nieves Gruñeiro, dean of the College of Arts and Sciences at East Stroudsburg, is watching for another evolution: whether using AI will actually make the articulation process more objective and less vulnerable to people’s often-unsubstantiated biases about other colleges’ rigor and reputation.
Gruñeiro said she is eager to see how a platform like CourseWise might start to “dismantle the preconceived notions and curricular misalignments” higher ed experiences. It’s “a way to really enter into collaborative conversations that will move credit mobility forward.”
