Published on

'Questions' Matter More Than 'Answers' β€” South Korea's Education Dilemma in the AI Age

"AI only answers as much as you ask it."

On May 7, 2026, Koh Young-sun, president of the Korea Educational Development Institute (KEDI), opened an education reform conference with this single sentence as a diagnosis of South Korea's education system. In a country where nearly 70% of young people go to university, the system remains locked in what Koh calls an "1980s mass-production model" β€” one designed to train people to find correct answers quickly. In an era where AI already performs that function better than humans, the question of what education is actually for has never been more urgent.

Table of Contents

  1. The Limits of Answer-Seeking Education β€” KEDI's Diagnosis
  2. The 1.4 Trillion Won AI Talent Plan: Promises and Problems
  3. Why Seoul National University's AI Expansion Was Rejected
  4. Vocational Schools and the AI Transformation Experiment
  5. Where Does Question-Based Education Begin?

1. The Limits of Answer-Seeking Education β€” KEDI's Diagnosis

Koh defined South Korea's biggest educational problem as its persistent focus on the ability to find correct answers. This is not a philosophical complaint β€” it is a structural reality in the age of generative AI.

Systems like ChatGPT, Claude, and Gemini already outperform humans at most "answer-finding" tasks: passing bar exams, clearing medical licensing tests, solving coding problems. The skills Korean students have spent decades training for are being automated.

So what remains distinctly human? Koh's answer is clear: "creating questions that have not yet been asked." AI is exceptional at answering questions given to it. But deciding which questions to ask, understanding why some questions matter more than others, and defining problems before solutions β€” these remain in the human domain.

South Korea's college entrance exam (CSAT/Suneung) and its multiple-choice-heavy assessment culture are optimized to test answer-finding speed. That is the system KEDI is asking to change.

2. The 1.4 Trillion Won AI Talent Plan: Promises and Problems

The South Korean government announced its "AI Talent Development Plan for All" in late 2025 β€” a β‚©1.4 trillion (approximately $1 billion USD) investment to train 11,000 high-level AI specialists and strengthen AI education from elementary school through graduate programs.

The plan is detailed:

  • AI Education Support Centers launched at three regional education offices starting 2026, expanding to all 17 districts nationwide by 2028
  • AI-focused schools to grow from 730 today to 2,000 by 2028
  • Smart science labs in every elementary, middle, and high school by 2027 (currently at 60%)
  • Seven new AI-specialized vocational high schools designated annually through 2030

The ambition is clear. But critics β€” including a Seoul Economic Daily analysis β€” warn that the plan is too narrowly focused on technical skill production. "In the rush to produce talent quickly, the fundamental question of what kind of thinkers, citizens, and learners the AI era requires has been left unanswered." This is precisely the gap that KEDI's call for "question-based education" is trying to fill.

3. Why Seoul National University's AI Expansion Was Rejected

On May 9, 2026, the Seoul Economic Daily reported that Seoul National University's application to expand its AI-related department capacity had been rejected. This directly contradicts the government's stated priority of building world-class AI talent.

The reason: South Korea's rigid university enrollment quota system. The Ministry of Education tightly controls the total number of students each university can admit. Expanding one department requires cutting another. The goal of producing world-leading AI researchers collided with a regulatory framework that has not meaningfully changed in decades.

The symbolism is hard to miss. While the government invests billions in AI education, the actual gateway β€” institutional capacity β€” remains gated by a system built for a different era. Policy ambition and institutional reality are on a collision course.

4. Vocational Schools and the AI Transformation Experiment

Beyond the policy debate, changes are happening at the classroom level. The Gyeonggi Provincial Office of Education selected 19 schools for its 2026 "Vocational High School AI Capability Enhancement Project," designed to help students develop practical AI skills for field deployment β€” not just theory.

Meanwhile, Seoul's free digital learning platform Seoul Learn surpassed 40,000 users as of May 2026. The platform provides AI-powered personalized learning content to low-income students at no cost β€” a concrete attempt to prevent digital education from becoming another inequality driver.

These ground-level experiments show what change looks like when it escapes the bureaucratic bottlenecks. Small-scale, specific, and measurable.

5. Where Does Question-Based Education Begin?

South Korea's paradox is visible: investment in AI education is world-class, but whether that education produces the right kind of learners for the AI age remains unresolved.

If KEDI's diagnosis is right β€” that the ability to ask questions matters more than the ability to answer them β€” then the assessment system is the first thing that needs to change. You cannot measure question-generating ability with multiple choice. Standardized tests cannot capture the capacity to identify which problems are worth solving or to imagine questions that do not yet exist.

Change is structurally difficult. The rejection of SNU's AI expansion illustrates how solid the barriers are. But the clearer the need for change becomes, the harder it is to justify maintaining the status quo.

Sources

Discuss on Twitter β€’ View on GitHub
'Questions' Matter More Than 'Answers' β€” South Korea's Education Dilemma in the AI Age | MINSSAM.COM