From 2 to 66 Scoring 90 or Higher... AI Gets a Perfect Score on Tax Accountant Exam
Data Released by Democratic Party of Korea Lawmaker Hwang Jeong-ah
AI Problem-Solving Experiment on Tax Accountant and LEET Exams
'Claude Opus 4.7' Scores Perfectly in Introduction to Accounting
The number of candidates scoring 90 or higher in Introduction to Accounting on this year’s first-stage tax accountant exam surged 33-fold from the previous year. Meanwhile, the latest artificial intelligence (AI) model released just before the exam scored a perfect mark after solving the actual test questions.
According to data released on Nov. 20 by Hwang Jeong-ah, a Democratic Party of Korea lawmaker on the National Assembly’s Science, ICT, Broadcasting and Communications Committee, Anthropic’s U.S.-based AI model Claude scored 100 points after correctly answering all 40 questions in Introduction to Accounting when given the questions from the 63rd first-stage tax accountant exam, held on April 25. Claude Opus 4.7 achieved the perfect score.
Opus 4.7 was the latest model available at the time, having been released on April 16, nine days before the exam.
Han, who passed the certified tax accountant exam with the highest score after using artificial intelligence (AI) to cheat, appears at the Public Crimes Investigation Unit of the Seoul Metropolitan Police Agency's Metropolitan Investigation Unit in Mapo-gu, Seoul, on the 17th as a suspect. Yonhap News.
View original imageHwang’s office provided the AI only with images of the test paper, without supplying answers or explanations. The responses were graded against the final answer key published by the Human Resources Development Service of Korea. Earlier models performed relatively poorly: Claude Sonnet 4.5, released last year, scored 62.5 points, while Haiku 4.5 scored 25. Opus 4.5 scored 75. By contrast, Opus 4.6 and Sonnet 4.6, both released in February this year, each scored a perfect mark.
A similar trend emerged in the actual tax accountant exam, with a sharp rise in high scores in Introduction to Accounting. According to the Human Resources Development Service of Korea, the number of candidates who scored 90 or higher in the subject jumped to 66 this year, from one in 2024 and two last year. The highest score also rose to 100 this year, from 90 in 2024 and 95 last year. One candidate scored a perfect mark in Introduction to Accounting this year.
By contrast, in Introduction to Tax Law, only one candidate scored 90 or higher this year, and that candidate earned a perfect mark. The number of high scorers increased notably only in Introduction to Accounting, the subject in which AI scored perfectly using only the test paper, without any additional materials.
A similar result was seen in the Law School Admission Test (LEET), the entrance exam for law schools.
In the 2027 LEET, held on July 19, an AI model called Fable 5, released 40 days before the exam, answered all 30 questions in Language Comprehension and 39 of the 40 questions in Logical Reasoning using only the test paper. After being trained on past exam questions from the previous three years and instructed to solve the questions in separate batches, it scored a perfect mark on all 40 Logical Reasoning questions as well.
AI models released after the exams performed even better. An analysis of nine AI models released after the tax accountant exam found that all scored 90 or higher in Introduction to Accounting, and six of them earned perfect marks. In particular, Fable 5.2, released this month, scored perfectly in both Introduction to Accounting and Introduction to Tax Law using only the test papers, without any additional training, and answered all 70 LEET questions correctly. The latest model completed all 40 questions in Introduction to Accounting in about one minute and 30 seconds.
Hot Picks Today
"Two Spend 500,000 Won in a Flash, Yet Wait Two Hours?" ... 'Dreamland' Shunned by Kids and Foreign Tourists
- The Era of Getting 200 Million Won After 30 Years Is Over... A Japanese Bank's "Perpetual Department Head" Who Just Clocked In Cries Out
- Spain Hit by Its Worst Tragedy... Lee Kangin, Moreno and Real Madrid Join Soccer World in Mourning
- They Said, “Make It Your Home for Life” - Then Suddenly Kicked 200 Passengers Off... What Happened to the Around-the-World Cruise?
- "130 Rice Bowls Sold Out in a Day"... 'Unfiltered Face' Adds 8 Million Won to Night Market Sales: China's Bungeunwoo
“It is difficult to dismiss as coincidence the fact that the number of high scorers increased more than 30-fold as AI models’ problem-solving capabilities improved,” Hwang said. “We need to investigate all candidates for statistical anomalies among high scorers and clearly determine whether any cheating took place.” She added, “AI has reached a level where it can score perfectly in tax law and the LEET just seven months after the exams, and its performance will continue to improve. We must develop measures to ensure fairness, including reviewing the monitoring and supervision measures applied to this year’s College Scholastic Ability Test and expanding them to all national qualification exams.”
© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.