The rising use of artificial intelligence in international student recruitment poses several risks to the higher education sector and is part of a concerning “move fast and break things” mindset, according to experts.
A wide range of AI products have been rolled out by agents and companies at different stages of the recruitment process, including chatbots that can communicate with students around the world 24 hours a day and help them narrow down their search for a university place.
Edysor, an education consultancy firm based in India, offers “real, human-like AI voices” – in a variety of languages, tones and paces – to speak to students. It says it can handle thousands of calls simultaneously, reducing wait times and improving the applicant experience during peak admissions periods.
Similarly, INTO University Partnerships’ AI admissions process claims to be able to cut down application processing time from weeks to just hours for students applying to leading universities in the US, UK and Australia.
ADVERTISEMENT
Andy Fawcett, the company’s chief technology officer and executive vice-president of admissions, said INTO does not yet use generative AI – which he said should only be deployed carefully, given students’ futures are at stake.
Instead, it uses large datasets to identify students most likely to enrol, or those who may need more support, and delivers them a 24-7 service with much lower levels of human intervention.
ADVERTISEMENT
Thomas Lancaster, a senior teaching fellow in computing at Imperial College London, said the big draw for universities in using tools like these was the ability to present tailored information to prospective students and increase conversions.
“The biggest problem comes when information is wrong, the AI hallucinates, or it misrepresents the university experience the student will have.
“When every student gets different information, it’s very hard to keep track of what the students are being told and they may be being recruited under false pretences. That leads to reputational risks and financial risks if students have been a mis-sold a place.”
INSELECT, a fully automated platform, boasts of “no humans, no hassle” as part of its “effortless and scalable admissions” process for universities. Its website claims it can help any university “select students like Harvard and Oxford” and revolutionise admissions “like VAR [Video Assisted Refereeing] transformed football” for overworked staff.
Its founder Dmitri Nersesyan told Times Higher Education that “thoughtful AI deployment” can help transform it from a tool into a trusted partner in admissions and advising.
But Lancaster warned that human oversight is always needed to ensure students can cope with the academic demands and rigour of the courses they’re joining, and have the language ability needed.
“There is also a risk of students finding the whole process very impersonal, which could end up doing more harm to successful recruitment than good,” he added.
ApplyBoard launched an AI adviser for studying abroad last year. “Abbie” can talk with students and answer their questions in real time about providers around the world, including application processes and courses.
ADVERTISEMENT
Haitham Amar, vice-president of AI strategy at the admissions company, said all AI features were developed using anonymised data, rigorously tested against objective benchmarks, and integrated into human-in-the-loop systems.
“Our goal is to responsibly leverage AI to improve outcomes, not compromise them.”
ADVERTISEMENT
Carissa Véliz, associate professor at the University of Oxford, said AI systems, which are typically vague around data collection, can trigger privacy risks.
“The privacy concern is particularly worrisome because prospective students might not have a choice to opt out. If they have to agree to subject themselves to this data collection or else not apply for an education, that’s not much of a choice.”
With a lot of “snake oil” in the area, Véliz said there was enough evidence to show that AI is not particularly reliable or trustworthy in selecting the best students, often displaying sexist and racist tendencies.
“It’s the responsibility of universities to do their best to admit students on the basis of fair criteria and processes, and that moral responsibility cannot be outsourced to a machine. The devil, as always, is in the details of how exactly we use AI and what kind of AI,” she said.
Students are also using AI platforms to write applications, support them in preparing for tests, and help them conduct interviews.
This poses an additional risk when the two halves of the automated recruitment process meet, according to Cato Rolea, assistant director for digital transformation at Ecctis, a UK firm that evaluates international qualifications.
“If you keep the human out of the loop, then you just have AI that’s verifying AI, so then you will end up with very bad potential applicants, but also misaligned expectations.”
Another Indian firm, iSchoolConnect, provides students with personalised AI recommendations from a list of over 300,000 courses and the ability to apply to multiple universities at once.
Rolea said practices such as this can lead to university staff having to deal with “massive bottlenecks” of applications, many of which turn out to be “nonsensical”.
Lauren Goodlad, distinguished professor of humanities at Rutgers, the State University of New Jersey, said safe tools can be created to help students but that many tech start-ups try to “cut corners” and “ride the wave of enthusiasm for generative AI”.
“The issue is that university administrators, like everybody else, are not especially knowledgeable about what works and what doesn’t.”
Echoing the famous mantra of Meta’s Mark Zuckerberg, she added: “It’s a move fast and break things free-for-all.”
ADVERTISEMENT
All firms mentioned were approached for comment.
patrick.jack
