TL;DR:

South Korea is attempting the most ambitious AI-in-education rollout on Earth — AI Digital Textbooks (AIDT) for every student, one device per child, 10G wireless in every school. But in 2026, the policy is already retreating. Teachers are drowning in admin work, the hagwon industrial complex is adapting faster than schools can reform, and the country with the world's lowest birth rate (0.72) is watching its teaching profession collapse. This is the story of a nation trying to fix education with technology while ignoring the humans who make education work.

The Final Chapter: Where Korea's AI Ambitions Meet Reality

This is Part 6 — the finale — of my Korea's AI Playbook series. We've covered national strategy, K-pop, bio-pharma, and more. But education is where it all comes together. Or falls apart.

South Korea spends more per student on private education than almost any country on Earth. Korean parents collectively pour over ₩26 trillion (~$19 billion) per year into hagwons — private cram schools that form a shadow education system more powerful than the public one. The average Korean high school student studies until 10 or 11 PM, and the college entrance exam (수능, Suneung) is so important that flight paths are rerouted and police escort late students to test centers.

Into this pressure cooker, the Korean government is now injecting AI. The question isn't whether technology can improve education — it obviously can. The question is whether a system this stressed, this competitive, and this human-dependent can absorb a technological transformation without breaking.

AIDT: The Most Ambitious AI Textbook Project on Earth

The Vision

In 2023, Korea's Ministry of Education unveiled its "Digital-Based Education Innovation Plan" — and at its center was the AI Digital Textbook, or AIDT (AI 디지털교과서). The pitch was bold:

Phase 1 (2025): Pilot deployment in select subjects
Phase 2 (2026): Full rollout — Grades 5-6, Middle School Year 2
→ 1 device per student (1인 1기기)
→ 10G wireless infrastructure in all participating schools
→ 1,900 AI pilot schools (AI 선도학교)
Phase 3 (2027+): Expansion to all grades and subjects

Core Features:
- Real-time learning analytics per student
- AI-powered adaptive content (difficulty adjustment)
- Teacher dashboard for class-wide insights
- 'K-Education AI' brand for global export

The ambition was staggering. Korea wanted to be the first country to replace traditional textbooks at national scale with AI-powered adaptive learning systems. Not as a supplement — as the primary medium of instruction.

The 2026 Retreat

Then reality hit.

By early 2026, something shifted. Education news outlet EduMorning published a devastating analysis:

"AIDT was once the centerpiece of Korean education innovation. But in the Ministry of Education's 2026 work report and budget proposal, it has been effectively pushed out of the policy center. The Ministry is redefining AI not as a textbook but as a 'supplementary educational tool,' transitioning from a speed-first AI education policy to a phased, distributed approach."

The retreat wasn't random. It was forced by a cascade of practical problems that anyone who's worked in a school could have predicted:

The Device Management Nightmare. Teachers reported spending their pre-class preparation time opening charging cabinets, troubleshooting broken tablets, and resolving network connectivity issues. One teacher's account captures it perfectly: "Without dedicated 'tech manager' staff, when the AI freezes during class, the teacher becomes a repair technician."

The Infrastructure Gap. Korea's cities have world-class connectivity. Rural and small-town schools? Not so much. The urban-rural digital divide that Korea thought it had solved with broadband in the 2000s is re-emerging with AI infrastructure.

The Teacher Readiness Problem. The Ministry doubled elementary school IT education hours to 68 per year. But who teaches those hours? Many elementary teachers entered the profession to teach children, not to debug WiFi routers and calibrate learning algorithms.

What AIDT Gets Right (In Theory)

Let me be fair to the vision, because the underlying insight is sound. Korea's education system has a personalization problem that AI is uniquely positioned to solve.

In a typical Korean classroom of 25-30 students, the range of ability is enormous. The teacher teaches to the middle, advanced students are bored, struggling students fall behind, and everyone goes to hagwon afterward to get the personalized instruction they didn't get in school. AIDT's promise — real-time adaptive content that meets each student where they are — could theoretically break this cycle.

Traditional Classroom:
→ 1 teacher, 25-30 students, 1 pace
→ Advanced students: understimulated → bored
→ Struggling students: overwhelmed → disengaged
→ Result: Everyone goes to hagwon anyway

AIDT Promise:
→ AI adapts difficulty per student in real-time
→ Teacher sees dashboard of class understanding
→ Intervention targets students who actually need it
→ Result (theoretical): Less need for hagwon supplementation

The problem isn't the technology. It's that you can't deploy technology into a system that's already collapsing under its own weight and expect it to be the thing that saves it.

The Hagwon Industrial Complex: Korea's Shadow Education System

By the Numbers

To understand why AI in Korean education is so fraught, you need to understand hagwons.

Korea's private tutoring market is not a supplement to public education — it's a parallel system that in many ways is the real education system. The numbers are staggering:

Annual private education spending: ₩27.1 trillion (~$19.5B)
→ Per-student monthly average: ₩43.4만 (~$312)
→ Seoul average: significantly higher

Student participation rate: 78.3% of all K-12 students
→ Elementary: 85.9%
→ Middle school: 76.2%
→ High school: 68.8%

Number of registered hagwons: ~75,000+
Hagwon industry employees: ~300,000+

Daechi-dong (대치동, Gangnam):
→ Korea's hagwon capital
→ ~1,200 hagwons in a single neighborhood
→ Annual revenue estimated at ₩20 trillion+

Here's the thing about hagwons that outsiders miss: they're not all evil cram factories. The best ones are genuinely excellent at personalized instruction. A top hagwon math teacher might earn ₩1-3 billion per year (yes, billion won — $700K to $2M+) because they're that effective at raising test scores. The hagwon industry has already been doing "adaptive learning" for decades — it's just humans doing it instead of algorithms.

AI Won't Kill Hagwons. Hagwons Will Adopt AI.

This is the part the Ministry of Education seems to have underestimated. When you introduce AI tutoring into public schools, you're not competing with hagwons — you're giving them a new product to sell.

The moment AIDT was announced, hagwon chains started developing their own AI tutoring platforms. Major edutech companies like Megastudy (메가스터디), Etoos (이투스), and Daekyo (대교) have been pouring investment into AI-powered adaptive learning. These are companies with decades of data on what Korean students struggle with, taught by instructors who are already masters of the test-prep game.

The structural problem is incentives. Public school teachers are evaluated on process compliance — did you follow the curriculum, did you submit the paperwork, did you use the mandated digital tools. Hagwon teachers are evaluated on outcomes — did the students' scores go up. When AI tools arrive in both settings, which system do you think will deploy them more effectively?

Neulbom Schools: The Childcare Crisis Meets the School System

What Are Neulbom Schools?

While AIDT targets how students learn, Neulbom Schools (늘봄학교) address a completely different crisis: who takes care of children while parents work.

Neulbom (늘봄, roughly "always spring") is the government's expanded after-school care program for elementary students. Launched in 2024, the plan is ambitious:

• 2026 target: Expand coverage to all students through Grade 3
• Integration with the early childhood education-care unification (유보통합) for ages 4-5
• Connect school-based care with community resources ("온동네 초등돌봄")
• Provide transportation support staff for safe commutes home

The policy exists because Korea has a dual crisis: the world's lowest birth rate (0.72 in 2024) and high dual-income household rates. Young couples cite childcare burden as the #1 reason for not having children. Neulbom is the government saying: "Have kids. We'll take care of them after school."

The Ground Truth

The vision sounds great in a press release. Here's what it looks like in a school.

A teacher who's also a parent published a blistering account in OhMyNews (December 2025) documenting the gap between policy and reality:

Space: Schools are already at capacity. Creating Neulbom classrooms means repurposing regular classrooms or eliminating specialty rooms (art, music, science labs). The government's parallel plan to open school facilities to the community means more outsiders in the building — raising security concerns for students.

Staffing: Neulbom care workers face poor pay and job instability. The hiring and management of Neulbom instructors becomes yet another administrative burden dumped on teachers who are already overwhelmed.

Identity Crisis: Teachers are asking a fundamental question that the policy doesn't answer: "Is a school an educational space or a community multipurpose center?"

Government Promise:
'Schools will provide comprehensive care, combining education,
childcare, culture, and community services.'

Teacher Reality:
'We can barely manage teaching. Now we're also running
a daycare, a community center, a facility management
operation, and an HR department for part-time instructors.
When do we prepare lessons?'

This isn't a technology problem. It's a labor problem disguised as a policy innovation.

The Teacher Crisis: Korea's Education System Is Losing Its Humans

A Profession in Collapse

Everything I've described so far — AIDT, Neulbom, college entrance reform — lands on the shoulders of teachers. And Korea's teachers are breaking.

The Seo-i-cho Incident and Its Aftermath. In July 2023, a first-year elementary teacher at Seo-i Elementary School in Seoul died by suicide. The incident exposed a systemic crisis of teacher abuse: parents filing malicious child abuse complaints against teachers for normal disciplinary actions, weaponizing Korea's child welfare laws to intimidate educators.

The tragedy sparked a national movement. Tens of thousands of teachers rallied. The government responded with measures:

1. Institutional complaint handling via education offices (교육지원청)
2. Unified school phone numbers (reduce direct harassment)
3. Strengthened penalties for parental misconduct
4. Expanded 'healing leave' for affected teachers
5. Proposal to record severe teacher-rights violations in student records

What Teachers Actually Want (But Didn't Get):
× Immediate legal representation for false abuse complaints
× Criminal penalties for malicious/frivolous reports
× Fundamental amendment to the Child Welfare Act
× Actual immunity from prosecution during investigation

The gap between what the government offered and what teachers need is a chasm. The institutional complaint-handling system sounds good until you realize the teacher still has to respond to the initial complaint and provide their account. The proposal to record violations in student records is a double-edged sword — it could deter some parents but also trigger more lawsuits from those who feel their children are being penalized.

The Demographic Collapse

Korea's birth rate of 0.72 — the lowest on Earth, and falling — is creating a structural crisis in teacher employment:

Birth Rate Trajectory:
2015: 1.24 → 2020: 0.84 → 2024: 0.72 → 2025 est: ~0.68

Impact on Schools:
→ Student population: 6M (2015) → ~4.5M (2026) → projected 3M (2035)
→ Teacher positions: shrinking proportionally
→ New teacher hiring: declining sharply
→ Teacher college applications: collapsing

Consequence:
→ Teaching becomes less attractive (fewer positions, harder to get hired)
→ Best graduates choose other careers
→ Quality of incoming teachers potentially declines
→ Remaining teachers bear heavier non-teaching burdens
→ More teachers burn out and leave
→ [Spiral continues]

Rural and small-town schools are being consolidated at an accelerating rate. When a school closes in a Korean rural community, it's often the final nail in that community's coffin — young families have no reason to stay, and the demographic spiral tightens.

The Administrative Burden

On top of emotional abuse from parents and existential demographic threat, Korean teachers are drowning in non-educational administrative work:

• School RE100: Managing solar panel installations on school rooftops
• Education Development Special Zones (교육발전특구): Coordinating with local government economic initiatives
• Neulbom management: Hiring, scheduling, and overseeing after-school care staff
• AIDT implementation: Device management, network troubleshooting, digital literacy training
• Endless 공문 (official documents): The bureaucratic paper chain that governs Korean public institutions

A veteran teacher's summary cuts to the bone: "Without a dramatic expansion of dedicated administrative staff, teachers will spend their time processing official documents instead of researching better lessons."

The 2028 College Entrance Overhaul: More Complexity, More Hagwons

What's Changing

As if teachers didn't have enough to deal with, Korea's college entrance system is undergoing its most significant restructuring in years. The 2022 Revised Curriculum takes effect for students entering high school in 2025, which means the 2028 Suneung (administered late 2027) will be fundamentally different.

Key changes:

• High School Credit System (고교학점제): Students choose more of their own courses, similar to a university model. Grades shift toward absolute evaluation (pass/fail thresholds) rather than pure relative ranking.
• Suneung restructuring: Subject categories and test formats are being redesigned to align with the credit system.
• Internal grades (내신) reform: How school grades are calculated and weighted is being overhauled.
• New university admissions patterns: Yonsei and Hanyang universities are introducing internal grade consideration even in the Suneung-based (정시) admissions track — a significant shift aimed at reducing the advantage of repeat test-takers (N수생).

The Complexity Trap

Here's the fundamental paradox of Korean education reform: every attempt to simplify the system makes it more complex, and every increase in complexity benefits those who can afford expert navigation.

The 2028 reforms create particular challenges for rural schools. The High School Credit System requires schools to offer a wide variety of courses. A large Seoul high school with 1,500 students can offer dozens of electives. A rural school with 200 students might struggle to staff more than the basics. The government's solution — online courses and inter-school collaboration — works on paper but introduces yet more technology dependence and logistical complexity.

Policy Goal: Reduce hagwon dependency through education reform
Actual Result: Create new categories of complexity that require expert guidance

Examples:
High School Credit System
→ 'Which courses should I take?' → Hagwon consulting service

Absolute Grading
→ 'How do universities interpret this?' → Hagwon admissions analysis

AI Digital Textbooks
→ 'How do I supplement the AI gaps?' → Hagwon AI tutoring platform

2028 Suneung Changes
→ 'What's the optimal test strategy?' → Hagwon test-prep revision

Net effect on hagwon industry revenue: ↑ (probably)

The Bigger Picture: Technology Can't Fix a Labor Crisis

What Korea Gets Right

Let me be clear about what's genuinely impressive here. Korea is one of the few countries attempting a systemic integration of AI into education rather than leaving it to individual teachers or schools. The infrastructure investment is real — 10G wireless in schools, one device per student, a national AI textbook platform. The ambition to create "K-Education AI" as a global export parallels the country's successful playbooks in broadband, smartphones, and K-content.

Korea also has structural advantages that most countries lack:

• Universal high-speed internet with near-100% penetration
• A culturally tech-forward population that adopts new technology rapidly
• Centralized education governance that can mandate nationwide changes
• An existing private sector (hagwons, edutech companies) that provides competitive pressure for innovation

What Korea Gets Wrong

The fatal flaw is treating education reform as a technology deployment problem when it's fundamentally a human capital problem.

Every AI textbook needs a teacher who understands how to use it. Every Neulbom classroom needs a care worker who's paid enough to stay. Every college entrance reform needs counselors who can guide students through the complexity. Every school needs administrators who handle the paperwork so teachers can teach.

Korea is deploying 21st-century technology into a system that can't adequately support its 20th-century workforce.

Where Korea IS investing:
✓ AI textbook platform development
✓ Device procurement (tablets, laptops)
✓ Network infrastructure (10G wireless)
✓ Pilot programs and showcase schools
✓ Policy announcements and branding

Where Korea is NOT investing enough:
✗ Teacher salaries and working conditions
✗ Administrative staff to reduce teacher burden
✗ Legal protections for teachers against abuse
✗ Neulbom care worker compensation
✗ Rural school capacity for credit system courses
✗ Mental health support for students AND teachers

The Hagwon Question

The elephant in every room of Korean education policy is hagwons. No government has successfully reduced Korean families' reliance on private tutoring, despite decades of attempts including outright bans (ruled unconstitutional in 2000), curfews (10 PM hagwon shutdown), and various "public education enhancement" initiatives.

AI doesn't change this dynamic — it intensifies it. If the public school AI system provides Level 5 personalization, hagwons will offer Level 10. If AIDT identifies a student's weak areas, the hagwon will sell the remediation. The technology is agnostic about equity; the market is not.

The only thing that could genuinely reduce hagwon dependency is reducing the stakes of college admissions — and that would require dismantling the social hierarchy that makes a degree from SKY (Seoul National, Korea, Yonsei) the gateway to elite employment. No government has shown the appetite for that fight.

What Happens Next

Korea's education system in 2026 sits at an inflection point that mirrors the country's broader relationship with AI (which we've explored throughout this series). The technology is real, the investment is substantial, the ambition is genuine — but the gap between policy announcements and ground-level reality is measured in human suffering.

A teacher quoted in OhMyNews said it best:

"What matters more than flashy AI devices is giving teachers the breathing room to use them. What must come before comprehensive childcare policies is securing safe spaces. What's more urgent than strong teacher-protection measures is an environment where teachers can focus solely on teaching."

The punchline is almost unbearably simple: teachers need to be happy for students to be happy. That's not a technology problem. It's a political choice.

Korea has proven, over and over, that it can execute ambitious national technology projects. Broadband, 5G, semiconductors, K-content — when Korea decides to build something, it builds it fast and at scale. But education isn't infrastructure you can deploy. It's a relationship between humans — teachers and students — that technology can enhance but never replace.

The AI Digital Textbook will arrive. Neulbom will expand. The 2028 Suneung will be administered. But whether any of it actually helps Korean students depends entirely on whether Korea invests as much in its teachers as it does in its tablets.

This is Part 6 of 6 in the Korea's AI Playbook series. Thanks for following along. Korea's AI story is far from over — but its education chapter may be the one that matters most.

Sources: OhMyNews (2025.12), EduMorning (2026), EduJin (2025), Korea Policy Briefing (2025.03), Ministry of Education Work Report (2025.12), various Korean education policy documents.