시험장까지 침투한 생성형 AI, 공정성 시험대에 오른 대한민국 교육 시스템 > K-wave Trends

Last May, for the first time in the history of official language proficiency tests in Korea, an attempt to cheat using AI glasses was captured. According to YBM, the organizer of the TOEIC in Korea, two different examinees were caught by the sharp eyes of proctors while attempting to take regular exams wearing smart glasses over the course of one month. The individuals caught were wearing either products already released overseas or models not yet available in Korea, suggesting that this was not merely a matter of curiosity, but a meticulously planned act of cheating. This incident serves as a declaration that traditional methods of controlling electronic devices are no longer sufficient to completely block cheating in official language exams.

The reason AI glasses are gaining attention as a new means of cheating is due to their lethal functionality. While these devices look identical to regular glasses and are difficult to identify with the naked eye, once the high-definition camera embedded in the frame captures the exam paper, generative AI immediately recognizes and analyzes the questions. The analyzed answers or hints are projected onto the inside of the lenses, allowing the examinee to check information in real-time as if reading a cheat sheet. In particular, smart glasses released recently can be operated via voice commands, and their multimodal capabilities allow them to interpret even complex foreign language passages in an instant, posing a severe risk to the fairness of exams.

The problem is that the speed at which these high-tech devices are being distributed is outpacing the response speed of exam management authorities. Following the official domestic release of smart glasses last month, a variety of product lines from major IT companies are expected to flood the market starting in the second half of this year. As the number of users increases, attempts to abuse these devices in exam halls are likely to become more sophisticated and organized. As cheating using AI glasses has already emerged as a social issue in universities abroad, it is urgent for domestic exam sites to strengthen security across the board and upgrade proctor training.

Education authorities and exam organizers have already begun discussions for proactive countermeasures. The YBM Korea TOEIC Committee is training proctors on how to identify the latest wearable devices and is strengthening post-exam management systems, such as analyzing answer similarity and verifying abnormal test-taking patterns. They are maintaining a firm stance that if cheating is detected, not only will the scores be invalidated, but the examinee will be disqualified for up to five years and face legal consequences. The Korea Institute for Curriculum and Evaluation (KICE) is also actively reviewing plans to codify regulations to fundamentally block the entry of AI glasses into national exams, such as the College Scholastic Ability Test (CSAT).

Meanwhile, in terms of operating the exam system, institutional improvements are being made to ease the burden on examinees. The Anti-Corruption and Civil Rights Commission has recommended linking systems so that language proficiency scores for public recruitment and qualification exams can be used for up to five years after being registered once. This measure aims to promote convenience for examinees by reducing the economic and time costs of retaking exams. In a situation where authorities must catch "two rabbits"—preventing cheating through technology and enhancing administrative convenience—they must establish strict security protocols while simultaneously creating a transparent system where examinees can demonstrate their skills in a fair competitive environment.