1. TL;DR — China AI Education at Teacher’s Day Eve
September 10, 2026 marks the 42nd Chinese Teacher’s Day (established by the Standing Committee of the National People’s Congress in 1985, observed annually on September 10). On the eve of Teacher’s Day 2026, China AI education shows a clear paradigm shift: from “single-point tool replacement” (smart speakers, grading apps) to a four-dimensional coordination of “Teacher Empowerment + Education LLMs + Policy Stack + Assessment Reform”.
The Ministry of Education’s August 2025 notice “Strengthening AI Education in Primary and Secondary Schools” sets the “2030 full K-12 AI education coverage” target. Combined with the September back-to-school dense deployment, the leading 7 education LLMs — iFlytek Spark, NetEase Ziyue, TAL Jiuzhang, Zhipu Qingyan (Education), ByteDance Doubao (Education), Alibaba Tongyi (Education), Baidu ERNIE (Education) — have formed an ecosystem across 4 application scenarios (smart lesson planning, intelligent grading, learning analytics, personalized tutoring).
Core conclusion: China AI education has completed a three-stage transition — Tool Enlightenment (2017–2022) → LLM Explosion (2023–2025) → Teacher Empowerment Coordination (2026–). September back-to-school is the key window for observing AI education landing. Schools should follow a 5-step path: Teacher Training → Subject Entry → Data Compliance → Assessment Integration → Home-School Coordination.
2. The Four-Dimensional Coordination
2.1 Policy Stack: MOE 2030 K-12 Full Coverage Target
In August 2025, the Ministry of Education (MOE) issued “Notice on Strengthening AI Education in Primary and Secondary Schools”, setting the “2030 full K-12 AI education coverage” target, accompanied by the “K-12 AI Education Curriculum Guide (Trial)” and “K-12 AI Education Teacher’s Handbook”. As the first year of the 15th Five-Year Plan, 2026 marks the entry into a “curriculum systematization + teacher scale-up + assessment scientificization + resource universalization” four-modernization coordination phase.
Key policy milestones:
- Aug 2025: MOE “Notice on Strengthening AI Education” (2030 full coverage)
- Jan 2026: MOE “AI + Education” special action second batch of typical case collection
- Mar 2026: MOE compulsory education quality monitoring first inclusion of AI literacy indicators
- Sep 2026: Fall semester opening — AI course coverage in pilot cities (Shanghai, Beijing, Shenzhen, Hangzhou, Chengdu, Suzhou) exceeds 60%
- Sep 2026: Ministry of Human Resources and Social Security + MOE jointly launch “AI Teacher” new occupation training program (first batch of 100,000 teachers)
2.2 Education LLMs: 7 Leading Players Ecosystem
China’s education LLM market has formed “big-tech education editions + edtech-native + regional vertical” three-tier players:
| Vendor | Education LLM | Core Scenarios | Deployment Scale (2026/9) |
|---|---|---|---|
| iFlytek | Spark Education | Smart grading + voice evaluation + dual-teacher class | 35,000 schools nationwide |
| NetEase Youdao | Ziyue (Confucius) | Personalized learning paths + K-12 Q&A | 12M+ K-12 users |
| TAL (Xueersi) | Jiuzhang (Nine Chapters) | Learning analytics + smart lesson planning + teacher assistant | 10,000+ TAL teachers |
| Zhipu AI | Qingyan (Education) | AI teaching assistant + thesis tutoring + subject Q&A | 500+ universities |
| ByteDance | Doubao (Education) | AI Q&A + photo-based problem solving + learning companion | 80M+ K-12 students |
| Alibaba | Tongyi (Education) | Smart classroom + teacher training + course generation | 50 regional “AI + Education” pilot cities |
| Baidu | ERNIE (Education) | ERNIE Bot Education + smart Q&A | 2,000+ universities and vocational schools |
2.3 Teacher Empowerment: From “AI Replaces Teachers” to “AI-Augmented Teachers”
Early AI education narratives often fell into the “AI replaces teachers” misconception. The 2026 new consensus is “AI-Augmented Teacher”: AI handles repetitive, mechanical work (grading, statistics, retrieval); teachers focus on emotional interaction, value guidance, and creative teaching.
4 typical teacher empowerment scenarios:
- Smart lesson planning: Teacher inputs teaching objectives + learning status data, AI generates lesson plan framework + courseware draft + exercise set, completing in 30 minutes what traditionally took 3 hours (iFlytek Spark, TAL Jiuzhang measured data).
- Intelligent grading: Objective questions auto-grading accuracy ≥99%; subjective questions (essays, oral English) AI-teacher grading correlation coefficient ≥0.85 (Shanghai Xuhui District 2025 pilot data).
- Learning analytics: Based on student answer data, AI auto-generates knowledge point mastery heatmaps + personalized error notebooks + weak link prediction (Zhipu Qingyan Education measured: 78% prediction accuracy).
- Personalized tutoring: Pushes adaptive learning paths based on learning profiles; per-student daily AI tutoring time capped at 30 minutes (MOE 2026 specification).
2.4 Assessment Reform: From “Score-Oriented” to “Competency-Literacy”
2026 AI education assessment reform shows three new trends:
- Process assessment share increase: Beijing, Shanghai, Shenzhen pilot schools raise process assessment share from 30% to 50%.
- AI literacy enters comprehensive quality assessment: Tsinghua High School, Shanghai High School, ECNU Second Affiliated High School 2026 autonomous admissions first include “AI project work” as bonus item (up to 20 points).
- Teacher AI literacy enters professional title review: Zhejiang, Guangdong 2026 teacher professional title review adds “AI teaching application capability” indicator.
3. September Back-to-School Dense Deployment
The September 2026 back-to-school season shows “policy density + product iteration + regional pilots + teacher training” four-line parallel deployment:
3.1 Policy Density
- MOE Sep 1: Releases “2026 Fall Semester AI Education Typical Case” collection notice (deadline Oct 31)
- MoHRSS + MOE Sep 3: Jointly launch “AI Teacher” new occupation training program (first batch 100,000 teachers; target 500,000 by end of 2027)
- MIIT + MOE Sep 5: “AI + Education” equipment national pilot (first batch 100 schools; Shenzhen Nanshan, Shanghai Xuhui, Beijing Haidian districts)
3.2 Product Iteration
- iFlytek Spark Education V3.5 (Sep 1): Adds “Teacher Digital Avatar” function
- NetEase Ziyue V2.0 (Sep 3): Full K-12 stage coverage + college entrance exam volunteer recommendation
- TAL Jiuzhang V4.0 (Sep 5): Multimodal courseware generation + Teacher AI Assistant APP
- Zhipu Qingyan Education V2.5 (Sep 7): University academic integrity detection + paper AI-assisted writing
3.3 Regional Pilots
- Shanghai: Xuhui District “AI Teaching Assistant” full coverage of 120 primary and secondary schools (Sep 1 opening)
- Beijing: Haidian District “AI + Education” equipment national pilot launched (first batch 30 schools)
- Shenzhen: Nanshan District “AI Teacher” full training + teaching-research AI assistant (50,000 teachers)
- Hangzhou: Gongshu District “AI Classroom” regularized (2 AI teaching assistants per school)
3.4 Teacher Training
- MOE “National Training Plan” 2026 AI Education Special (100,000 teachers, 5,000 principals)
- 7 education LLM vendors jointly launch “AI Teacher Empowerment Plan” (free training for 500,000 teachers)
- 100+ normal universities nationwide establish “AI Education” undergraduate majors (2026 first batch enrollment 5,000 students)
4. International Positioning
China AI education in 2026 presents a “policy leadership + model quantity leadership + scenario coverage leadership + ethics catch-up” four-dimensional landscape:
- Policy strength: MOE 2030 full coverage target is the world’s first national-level AI education K-12 full coverage timetable (US has no corresponding national target).
- Model quantity: 7 leading education LLMs + 200+ regional vertical models (languages, math, English, physics, chemistry, biology, politics, history, geography, music, PE, science-innovation, labor).
- Scenario coverage: Smart grading, smart lesson planning, learning analytics, personalized tutoring — 4 scenarios with the broadest global coverage.
- Ethics catch-up: Teacher data privacy + student data minimization + AI decision explainability + human teacher final decision rights — 4 principles first clarified by MOE July 2026 “Educational AI Application Ethics Guidelines”.
Global comparison (Sep 2026):
- USA: Education AI mainly tool-based (Khan Academy Khanmigo, Duolingo Max), no national AI education curriculum standards yet.
- UK: Department for Education May 2026 releases “AI Education Framework”, but landing mainly in higher education.
- South Korea: MOE 2025 releases AI digital textbooks, 2026 covers primary 3-4 + junior 1.
- EU: AI Act Aug 2026 enters full applicability phase, education AI must pass “high-risk” assessment.
5. 5-Step School Landing Path
For schools intending to advance AI education, recommended path: “Teacher Training → Subject Entry → Data Compliance → Assessment Integration → Home-School Coordination” 5 steps:
| Step | Time Window | Key Actions | Quantitative Metrics |
|---|---|---|---|
| 1. Teacher Training | First 2 weeks | All-staff AI literacy training + subject backbone in-depth training | 100% teachers pass basic assessment |
| 2. Subject Entry | Months 1-3 | Select 2-3 subjects (recommended: English, Math, Chinese) to pilot | Pilot subject coverage 60%+ |
| 3. Data Compliance | Full cycle | Student data localization + parental informed consent + third-party audit | 100% compliance filing |
| 4. Assessment Integration | Months 4-6 | Process assessment + AI literacy + teacher AI evaluation capability | Assessment reform plan lands |
| 5. Home-School Coordination | Full cycle | Parent AI training + home AI learning scenarios + communication mechanism | Parent satisfaction ≥80% |
6. 6 Defense Checklists
Schools advancing AI education need to guard against 6 major risks:
- Data privacy: Student data does not leave school / desensitized storage / minimal collection / third-party independent audit.
- Algorithm bias: Regularly audit AI output gender / region / class bias / multi-model cross-validation.
- Teacher subjectivity: AI decision-making authority belongs to teacher / teacher AI literacy into professional title review / reject “AI replaces teacher” narrative.
- Learning alienation: Per-student AI tutoring daily ≤30 min / prohibit AI ghostwriting homework / retain paper reading + outdoor activities.
- Commercial capture: Beware of education LLM vendor “data roundtrip + course lock-in + renewal” trap / government procurement compliance.
- Ethics bottom line: Prohibit AI for student psychological portrait discrimination / prohibit AI replacing teacher comments / retain human teacher final decision rights.
7. Key Terminology
| Term | Definition |
|---|---|
| Education LLM | Large language models specifically trained for K-12 / higher education / vocational scenarios, with deeper domain knowledge, curriculum alignment, and pedagogy understanding than general LLMs |
| AI-Augmented Teacher | 2026 new consensus: AI handles repetitive work (grading, statistics, retrieval); teacher focuses on emotional interaction, value guidance, creative teaching |
| Intelligent Grading | AI auto-grading objective questions + subjective questions (essays, oral English); correlation coefficient with teacher scoring ≥0.85 |
| Learning Analytics | Generated knowledge point mastery heatmaps, personalized error notebooks, weak link predictions based on student answer data |
| Process Assessment | Replaces traditional “score-only” approach; emphasizes participation, progress, collaboration in learning process |
| AI Literacy | Entered comprehensive quality assessment in 2026; includes AI tool use, AI ethics understanding, AI project creation |
| AI Teacher New Occupation | MoHRSS + MOE September 2026 jointly launched new occupation training program (first batch 100,000 teachers) |
| Educational AI Ethics Four Principles | MOE July 2026 “Educational AI Application Ethics Guidelines”: data privacy + algorithmic fairness + teacher subjectivity + student welfare |
8. FAQ
Q1: Will AI replace teachers? A: The 2026 new consensus is “AI-Augmented Teacher” rather than “AI replaces teacher”. AI handles repetitive, mechanical work (grading, statistics, retrieval); teachers focus on emotional interaction, value guidance, creative teaching. MOE July 2026 “Educational AI Application Ethics Guidelines” explicitly states “human teacher final decision rights”.
Q2: How is student data privacy protected in AI education? A: Four safeguards: ① Student data localization (does not leave school); ② Minimal collection (only necessary data); ③ Parental informed consent; ④ Third-party independent audit. MOE + CAC July 2026 jointly release “Educational AI Application Compliance Guidelines” detailing requirements.
Q3: How should schools select education LLM vendors? A: Three-step selection: ① Subject match (K-12 / higher ed / vocational scenario coverage); ② Data compliance (student data localization commitment); ③ Teacher experience (lesson planning / grading / learning analytics tool usability). Prioritize MOE 2026 typical case recommended vendors.
Q4: Is AI education equitable for rural schools? A: 2026 policies tilt rural: ① “National Training Plan” rural teacher AI training special; ② “AI + Education” equipment national pilot prioritizes central and western regions; ③ Education LLM vendors free training for 500,000 teachers (focus rural schools). But digital divide (bandwidth, devices) remains a bottleneck.
Q5: How can teachers avoid being “captured” by AI? A: Three lines of defense: ① Retain teacher decision rights (AI only as reference); ② Teacher AI literacy into professional title review (Zhejiang, Guangdong 2026 new rules); ③ Reject “AI replaces teacher” narrative, return to education essence (emotion, value, creation).
Q6: Will AI education exacerbate student “learning alienation”? A: Risks exist; need three defenses: ① Per-student AI tutoring daily ≤30 min (MOE 2026 spec); ② Prohibit AI ghostwriting homework; ③ Retain paper reading + outdoor activities + peer interaction. MOE + NHC September 2026 jointly release “Educational AI Application Health Protection Guidelines”.
9. References
Policy documents:
- MOE “Notice on Strengthening AI Education in Primary and Secondary Schools” (August 2025)
- MOE “K-12 AI Education Curriculum Guide (Trial)” (August 2025)
- MOE “Educational AI Application Ethics Guidelines” (July 2026)
- MoHRSS + MOE “AI Teacher New Occupation Training Program” (September 2026)
- MIIT + MOE “AI + Education Equipment National Pilot Plan” (September 2026)
Industry reports: 6. CAICT “Education LLM Development Research Report 2026” (June 2026) 7. iResearch “China K-12 AI Education Industry Research Report 2026” (March 2026) 8. 36Kr Research “EdTech 2026 Mid-Year Review” (July 2026) 9. Duojing Capital “2026 Education Industry AI Application Map” (August 2026)
Vendor dynamics: 10. iFlytek Spark Education V3.5 release (September 2026) 11. NetEase Youdao Ziyue V2.0 release (September 2026) 12. TAL Jiuzhang V4.0 release (September 2026) 13. Zhipu AI Qingyan Education V2.5 release (September 2026)
International comparison: 14. Khan Academy Khanmigo official blog (August 2026) 15. UK Department for Education “AI Education Framework” (May 2026) 16. South Korea MOE AI digital textbooks (2026) 17. EU AI Act education AI high-risk assessment guidelines (August 2026)
Media observations: 18. People’s Daily “AI Education Needs to Beware of ‘Tool Replacement’ Misconception” (September 2026) 19. China Education Daily “Teacher’s Day Eve: AI Education Four-Dimensional Coordination” (September 2026) 20. Caixin “7 Education LLM Players Ecosystem” (September 2026) 21. 21st Century Business Herald “AI Education Back-to-School Dense Deployment” (September 2026) 22. Yicai “MOE 2030 K-12 AI Full Coverage Roadmap” (September 2026)
Author: Juan Ru (boaoai.cn editor) Reviewer: Xiaohui Chang (boss) Publication date: September 9, 2026 (China Teacher’s Day Eve) Word count: Chinese ~3,800 chars / English ~1,500 words