AI in University Admissions: A Helpful Assistant or an Invisible Salesperson?
University admissions are becoming less like traditional open days and more like personalised digital funnels. Algorithms decide who sees advertisements for academic programmes, generative AI produces dozens of tailored email variations, chatbots provide around-the-clock guidance on deadlines and required documents, and interactive dashboards estimate an applicant’s chances of admission.
Dmytro Chumachenko
Candidate of Technical Sciences, Associate Professor, Associate Professor of the Department of Mathematical Modeling and Artificial Intelligence of the M.E. Zhukovsky National Aerospace University "Kharkiv Aviation Institute
Against the backdrop of demographic decline, Russia’s full-scale invasion of Ukraine, and growing competition for prospective students, such efficiency is understandably appealing. Modern AI systems can predict the likelihood that a prospective student will submit an application, personalise communication throughout the admissions process, and automate administrative support. Yet the central question remains: where is the line between assistance and manipulation?
Many universities across Europe and North America began exploring these technologies several years ago. The most convincing examples of AI adoption are those in which it removes administrative barriers rather than influencing applicants’ decisions.
One of the best-known examples comes from Georgia State University, which introduced the chatbot Pounce to support admitted students before enrolment. The chatbot reminded students about financial aid, registration deadlines, and required paperwork. According to the university, its implementation was associated with an almost 4% increase in enrolment and a 21.4% reduction in "summer melt"—the phenomenon in which admitted students ultimately fail to begin their studies.
Many universities across Europe and North America began exploring these technologies several years ago. The most convincing examples of AI adoption are those in which it removes administrative barriers rather than influencing applicants’ decisions.
One of the best-known examples comes from Georgia State University, which introduced the chatbot Pounce to support admitted students before enrolment. The chatbot reminded students about financial aid, registration deadlines, and required paperwork. According to the university, its implementation was associated with an almost 4% increase in enrolment and a 21.4% reduction in "summer melt"—the phenomenon in which admitted students ultimately fail to begin their studies.
In Spain, the chatbot Lola assisted incoming students at the University of Murcia and the Polytechnic University of Cartagena. According to its developers, Lola handled 38,708 questions from 4,609 users while achieving an accuracy rate of 91.67%. Meanwhile, the United Kingdom’s Leeds Beckett University went a step further. Its chatbot, Becky, not only provides prospective students with 24/7 admissions support but can also assess whether applicants meet programme requirements and generate preliminary admission recommendations.
Importantly, these early systems were not necessarily examples of generative AI like ChatGPT. Most relied on predefined conversational flows, verified knowledge bases, and relatively simple machine learning models. In university admissions, such predictability may actually be an advantage: it is better to receive a limited but accurate answer than a highly convincing AI hallucination about a scholarship that does not exist.
Importantly, these early systems were not necessarily examples of generative AI like ChatGPT. Most relied on predefined conversational flows, verified knowledge bases, and relatively simple machine learning models. In university admissions, such predictability may actually be an advantage: it is better to receive a limited but accurate answer than a highly convincing AI hallucination about a scholarship that does not exist.
Today, however, generative AI is increasingly being used not only to provide information but also to persuade prospective students. The Chronicle of Higher Education reported that South Dakota State University used ChatGPT to generate captions for promotional videos, while Emory University employed Photoshop’s generative editing tools to remove caution tape from a campus photograph used in recruitment materials. North Carolina State University also explored using generative AI to dramatically accelerate the production of personalised marketing messages.
These marketing experiments may appear harmless, yet they raise important questions about authenticity. Are prospective students seeing the university as it truly is, or an algorithmically optimised version designed to maximise appeal?
The risk of discrimination also exists even when AI is not directly involved in admission decisions. A predictive model may classify an applicant as "unlikely to enrol" and therefore withhold invitations to consultations, reminders, or information about financial aid. The algorithm does not need to reject an applicant outright. It is enough for one prospective student to feel welcomed while another becomes effectively invisible.
These marketing experiments may appear harmless, yet they raise important questions about authenticity. Are prospective students seeing the university as it truly is, or an algorithmically optimised version designed to maximise appeal?
The risk of discrimination also exists even when AI is not directly involved in admission decisions. A predictive model may classify an applicant as "unlikely to enrol" and therefore withhold invitations to consultations, reminders, or information about financial aid. The algorithm does not need to reject an applicant outright. It is enough for one prospective student to feel welcomed while another becomes effectively invisible.
This is why Ukrainian universities should establish clear and transparent principles for AI use. They should inform applicants whenever they interact with AI, rely exclusively on verified sources of information, provide a seamless option to connect with a human staff member, minimise the collection of personal data, clearly label synthetic or substantially edited images, and regularly evaluate AI systems for bias related to language, region, disability, or socioeconomic status.
These recommendations align with emerging international standards. The European Union’s AI Act classifies AI systems used to determine access to education as high-risk, while UNESCO emphasises privacy, fairness, transparency, and meaningful human oversight as fundamental principles for the responsible use of artificial intelligence in education.
These recommendations align with emerging international standards. The European Union’s AI Act classifies AI systems used to determine access to education as high-risk, while UNESCO emphasises privacy, fairness, transparency, and meaningful human oversight as fundamental principles for the responsible use of artificial intelligence in education.
A university is not an online marketplace, and a prospective student is not simply another marketing lead. The most valuable role for AI in university admissions is not to subtly steer people towards a particular choice, but to remove unnecessary bureaucratic obstacles while leaving life-changing educational decisions where they belong—in human hands.
05.08.2026
Did you like this article?