Artificial intelligence (AI) has scored close to perfect in the Korean language section of the University Academic Ability Test
According to the information and communication technology (ICT) industry, Marker AI, a domestic AI start-up, utilized open AI's AI model to undertake the Korean language test for the 2025 academic year. The "o1-preview" model achieved an impressive 97 points, equivalent to a first-grade performance.
The only instance in which 'o1-preview' erred was in question 8 within the Korean language segment. This question necessitated the understanding of two non-literary texts on modernization and the assessment of logical thinking by applying them to the scenarios presented.
This particular question marked the highest incorrect response rate of 81.5% among all the Korean language questions for the 2025 academic year. Marker AI elucidated that 'o1-preview' misinterpreted the context of fingerprints and views, failing to grasp the underlying purpose of the problem. Marker AI has been assessing the AI models' processing capabilities for the Korean language section of the CSAT over the past decade.
Recent Scores
The 'o1-mini' model, which successfully completed the Korean language section for the 2025 academic year, garnered a score of 78. Additionally, the 'gpt-4o' scored 75 points and was classified in the fourth grade.
