Editorial In an Era When AI Scores Top Marks, What Will the State Verify?

This year, the top passer of the tax accountant qualification exam confessed to cheating. Using a miniature camera worth 110,000 won, he photographed the exam papers and sent them out, then copied down answers provided by artificial intelligence (AI). It happened right under the invigilators' noses. He voluntarily turned himself in to police, and an investigation is underway. The state-certified number one that year has testified through his own words that the state's verification system has been breached.

Predictable responses are already emerging: strengthen supervision, block transmissions, increase penalties. All necessary measures. Yet the weight of this incident lies in the fact that this cannot be where it ends. Equipment gets smaller each year and models grow faster. There is no history of the pace of prevention outpacing the pace of circumvention. If approached solely as a security problem, the state will endlessly pour budgets into a race it is destined to lose.

The essence lies elsewhere. What this incident proves is that AI writes the top passer's answers in mere seconds. In other words, the state's number one has physically demonstrated that the value of the ability this exam has measured for decades—namely, the ability to accurately recall memorized knowledge—has collapsed. What he stole was not the answers but the questions themselves. What exactly was this exam verifying?

This question does not stop at the tax accountant exam alone. University entrance exams, written tests for corporate and public sector recruitment, and national qualifications that distinguish lawyers, accountants, and doctors all rest on the same premise: that measuring the amount of knowledge in a person's head reveals their competence. If AI undermines that premise, not just one gateway but three—entrance exams, recruitment, and qualifications—are shaken simultaneously. It has long been the case overseas that AI passes professional licensing exams with top-tier scores. What we have witnessed is merely the first scene of that reality walking into an exam room.

Therefore, what the state must now begin is not increasing surveillance cameras but redesigning what to verify. A pan-government discussion bringing educational curriculum, recruitment, and qualification systems to the same table is necessary. The agenda includes: what can be shifted to open-ended, practical evaluations premised on AI use; how to define and measure competencies that cannot be replaced by AI—namely, judgment, responsibility, and field expertise; and how to protect the trust of hundreds of thousands who obtained qualifications through existing methods during this transition. With the current structure where the Ministry of Education, Ministry of Employment and Labor, and various qualification authorities each hold a piece of the system, there is no way to answer this question.

The young man deceived the exam, but the exam has been deceiving the times. What the top passer's confession revealed is not one person's fraud but the state's void in knowing what to verify anymore. Punishment is for the courts to dispense. The question is for the state to receive. It is time to begin the discussion.