The National Education Commission, a presidential advisory body in South Korea, is reviewing a plan to introduce essay-type assessments in the college entrance exam and school records, with AI handling the initial grading. At a two-day public deliberative forum held August 29-30 in Gyeonggi Province, 180 participants - including students, parents, and civic group representatives - split into 24 groups to discuss the proposal ahead of the October announcement of the college entrance exam reform plan.
The central point of contention was grading. Nineteen of the 24 groups identified a "fair and transparent grading system" as a prerequisite for essay-type assessments.
Lee Min-seok, a senior research fellow at the NEC, outlined the proposed process: "The process would involve AI conducting the initial grading, followed by teacher re-evaluation, with teachers ultimately responsible for the final scores."
Participants generally agreed with the direction of introducing essay-type assessments, but skepticism about AI grading ran deep. One current teacher said, "Even now, teachers prioritize whether specific keywords or sentences are included in subjective answers to avoid complaints, rather than evaluating logical coherence. Introducing AI grading for the college entrance exam is hard to view positively."
A salaried worker in his 30s, identified only as Min, predicted a surge in legal challenges: "There will likely be far more administrative lawsuits challenging grading results than there are now."
Concerns about equity and the "black box" problem
Critics also argued that essay-type assessments could widen educational disparities based on parental income. Hong Soo-yeon, co-representative of the National Parents' Association, said, "Quick-acting parents will turn to private education providers to analyze grading criteria and examiners' intentions if essay-type admissions expand, further widening the education gap based on economic power."
Other participants said AI scores cannot be fully trusted and that a grading system ensuring objectivity and fairness should be established before essay-type assessments are introduced gradually.
The Korea Institute for Curriculum and Evaluation raised similar concerns in a recent report. It identified the "black box" issue - where AI's scoring rationale is unclear - as a fundamental limitation. The report also warned that in a system where AI grades first and teachers review, AI scores could act as a benchmark, leading teachers to formally approve AI scores without substantive review.
For educators weighing how to respond, the practical question is not whether AI will grade essays, but how to maintain meaningful oversight when it does. Teachers who understand how AI scoring systems work - including their tendency to reward keyword matching over logical coherence - will be better positioned to catch errors and push back on flawed scores. That's a skill set worth developing before the policy lands, not after. AI for Teachers Courses cover these classroom-facing tools and their limitations. For a broader look at how AI is reshaping assessment and instruction, see AI for Education.
Your membership also unlocks:
