Korea is considering a striking answer to a hard problem, using artificial intelligence to grade the essay questions it wants to add to the national college entrance exam. As the government prepares to reshape the test known as the Suneung, officials say AI is on the table as a way to score written answers fairly.
The idea sits inside a larger overhaul being drafted by the National Education Commission, which is gathering public opinion now and aims to finalize a blueprint by March 2027. The plan would add essay and written-response questions across every subject and move the exam toward absolute grading.
Why the machine is being considered
The motive is fairness. Scoring essays by hand invites doubts about consistency in a country where a few points can decide which university a student attends. A senior official said the biggest public concern is fairness and that the commission is reviewing the use of AI as an alternative approach to human marking.
That concern is not abstract. The Suneung is among the most consequential days in a young Korean's life, and any hint that two graders might score the same essay differently could undermine trust in the whole exercise. AI is being floated as a way to apply one standard uniformly.
A break from filling in bubbles
The reform would be a sharp departure from how the test works today. The Suneung is built on five-option multiple-choice questions, and under the current relative grading only the top 4 percent of takers earn the highest mark, a system that rewards ranking students against one another rather than measuring what each can do.
The commission's chairman, Cha Jeong-in, framed the change as a matter of principle, saying multiple-choice questions are ultimately just about choosing answers provided by someone else. Written answers, in that view, ask students to produce ideas rather than recognize them.
A president pushing the change
The effort has backing at the top. President Lee Jae-myung has argued that the current format is unsuitable for the age of AI, and he framed the stakes bluntly, saying the country should not sacrifice children's education for the sake of adults' easy, convenient evaluations. The line casts multiple-choice testing as a convenience for graders rather than a service to students.
The push extends beyond grading. The Ministry of Education has separately announced plans to use AI-generated English passages in mock exams in 2028, a sign that the technology is being woven into the testing system from more than one direction.
The doubts that remain
Not everyone is convinced a machine settles the fairness question. The Korean Teachers and Education Workers Union cautioned that the objectivity of AI depends entirely on how the system is designed, and urged the government to consult teachers, students and parents before locking in the reforms.
The warning points to the paradox at the center of the plan. AI is being enlisted to remove human bias from grading, yet the systems themselves are built by people and trained on their choices. Whether Korea can trust software with a test this important is a question the country will have to answer before the first essay is ever scored.