🇰🇷South Korea·AI News·19 Nov 2026·via 서울경제

    8 in 10 Koreans Trust Top Exam Scorers Less After AI Cheating Case

    Following a case where the top performer on South Korea's national tax accountant exam cheated using artificial intelligence, a PMI survey found that nearly 80% of South Koreans now trust top exam scorers less. The "Survey on Perceptions of Exam Fairness in the AI Era," released on the 18th, gathered insights from 1,000 adults nationwide.

    The integrity of national qualification exams in South Korea has come under scrutiny after a specific incident of AI-assisted cheating. A survey conducted by PMI on the 17th revealed that 79.5% of respondents aged 20 to 59 reported a decline in their trust in the scores of high-performing exam takers. This broad erosion of confidence follows widespread awareness of the cheating case, with 89.1% of those surveyed knowing about it, and over half knowing specific details.

    The survey quantified the shift in public perception, noting that 40.1% of respondents felt their trust had fallen sharply, and an additional 39.4% said it had fallen somewhat. Only a small fraction, 8.4%, indicated no change in their trust. This suggests that the reported AI cheating incident has had a tangible impact on public confidence in the fairness and reliability of exam results.

    In response to concerns about AI-assisted cheating, a significant 77.0% of respondents supported tighter management and supervision at exam sites. However, there were differing views on the intensity of these measures. While 46.7% advocated for aggressive strengthening, including detailed searches, radio signal detection, and closed-circuit television monitoring, even with privacy concerns, another 30.3% preferred a more limited approach to minimize privacy intrusion. The preference for aggressive tightening notably increased with age, reaching 58.0% among those in their 50s compared to 36.4% among those in their 20s.

    When asked about the most critical first step to prevent AI-assisted cheating, 33.7% of respondents identified the introduction of detection equipment, such as radio signal detectors. Other proposed solutions included tougher penalties and testing bans (15.9%), redesigning assessments to inherently make cheating more difficult (13.5%), and stricter checks on electronic devices and personal belongings at entry (13.2%).

    Beyond immediate prevention, the survey explored how exam and assessment systems should evolve with the widespread integration of AI. A majority of 57.0% of respondents believed that assessment methods should be revised or should reflect the ability to use AI. This group included 27.6% who suggested restricting AI use while focusing assessments on reasoning and problem-solving, 17.3% who thought some exams should specifically test AI proficiency, and 12.1% who proposed making AI proficiency a main assessment criterion. Conversely, 37.0% advocated for maintaining existing assessment methods while strengthening supervision against AI-assisted cheating, indicating a divided public opinion on the future direction of evaluations.

    Cho Min-hee, chief executive of PMI, stated that the survey confirmed AI-assisted cheating has moved beyond isolated incidents to become a fundamental question of trust in exam scores and assessment results. The findings underscore a growing need for discussions on management systems to ensure exam fairness, alongside a re-evaluation of how actual competence is assessed and verified.

    Share this article

    Original reporting by 서울경제We don't republish, read the full story →

    Related reading

    6 stories