AI cheating cases highlight need for updated policies

After cases of collective student cheating using the generative AI program ChatGPT were uncovered at Korea University and Yonsei University, similar incidents have been confirmed at other universities, including Seoul National University. Unlike the large-scale online exams at Korea University and Yonsei University, the cases at Seoul National University involved small in-person tests. AI-assisted cheating is spreading regardless of whether exams are online or in-person.

At Seoul National University, a recent midterm for the general education course “Statistics Lab” revealed that some students used ChatGPT to complete their answers. The course, which has about 30 students, is held in-person. A similar suspicion arose in December of last year, but the case was closed without disciplinary action due to insufficient evidence. In June, during an in-person exam for a major course at Seoul Women’s University, some students were caught using AI to answer descriptive questions and received zero points.

Cyber universities, where all classes and exams are conducted online, report that AI-assisted cheating is so widespread that students say “honesty is a disadvantage.” Despite recurring issues since ChatGPT was released three years ago, domestic universities remain largely unprepared.

Criticism over AI cheating on campuses has focused on students’ lack of ethical awareness. However, it is also necessary to examine whether the education system has adapted to the revolutionary changes AI brings. Major overseas universities have moved away from assessment methods focused solely on knowledge transmission and correct answers. They are developing learning and evaluation strategies suited to the AI era and providing clear guidelines on the extent to which AI use is permitted.

In contrast, very few domestic universities have established AI guidelines. Many have increased large online courses, which are particularly vulnerable to AI-assisted cheating, due to financial difficulties caused by prolonged tuition freezes. For these institutions, creating AI guidelines is considered a luxury.

Even financially well-resourced primary and secondary schools are lagging. Among 17 metropolitan and provincial education offices, seven have no guidelines for AI use. Those that have issued guidelines generally provide only broad principles, such as requiring ethical use education, making them largely ineffective.

Evaluation results in middle and high schools directly affect university admissions, making fairness a particularly sensitive issue. Education and assessment methods must be updated to align with a generation for whom AI is a daily tool, the so-called “ChatGPT generation.” In primary and secondary schools, the Ministry of Education should provide concrete guidelines for AI use.