AI Education in China’s Primary and Secondary Schools

How is China turning national direction on AI education into local curricula and real classroom practice?

5 claims79 direct citations79 original sourcesUpdated 2026-10-05

Written and maintained by Robin · Updated 2026-10-05

This answer synthesizes public sources. Check the evidence and limits before applying it. Review the evidence

Citation text

Robin. How is China turning national direction on AI education into local curricula and real classroom practice?. 2026-10-05.
AI education in Chinese schools continues to move from common requirements toward regional plans, school-based professional learning, and student projects. Dongguan has placed AI education in its development plan, Beijing is building future-learning spaces, and Shanghai's Songjiang District has set a school-based training target. Plans, reach, and participation still need to be connected to stable hours, teacher use, student process artifacts, and comparable outcomes.
https://edu.hhhh.life/en/topic/china-path/#answer

LIVING ANSWER

Follow the question. Read the answer directly.

Start with the claims, then open the evidence when you need it.

QUESTION 1

How does national direction become local curriculum?

National documents have moved AI education from an initiative to a local curriculum responsibility. Local authorities must still translate that direction into curriculum guides, class hours, resources, and assessment systems.

After the national action plan specified stage-based goals, content, and class hours, local authorities began taking on verifiable responsibility for course delivery. Jiangsu is expanding grades 1-12 provision, Dongguan has included an AI-education pilot plan in its education strategy, and Shanghai's Songjiang District has set a school-based training target. Implementation has to appear in timetables, assigned teachers, resource use, teacher practice, and delivery at each school. [01][02][03] plus 17

20 sources for this claim Supports 2 · Illustrates 5 · Tracks 13 View all ↓Collapse ↑

National general-education materials covering four educational stages and public courses provide a base layer of resources. Jiangsu has also assembled more than 130 national-level courses and nearly 800 hours of structured resources, while providing rural schools with cloud classrooms and technical support. The volume of resources can lower the barrier to offering courses, but actual use, the quality of updates, and follow-up teacher training still need to be disclosed over time. [21][01][22] plus 13

16 sources for this claim Supports 2 · Qualifies 1 · Illustrates 10 · Tracks 3 View all ↓Collapse ↑

QUESTION 2

How can we tell whether AI education has reached real classrooms?

The minimum evidence is stable class time and teacher capacity. Stronger evidence also includes student work, classroom assessment, sustained use, and quality differences among schools.

Beijing schools are incorporating agent building, robotics experiments, and interdisciplinary work into age-sequenced AI literacy classes, with a stated minimum of eight hours per year. A Mudanjiang Chinese-language model workshop and new embodied-intelligence and AI-agent competition categories broaden the task formats. Demonstrations, workshops, and competitions show a direction of practice, while routine timetables, participation, and learning changes still need disclosure. [36][37][38] plus 8

11 sources for this claim Supports 1 · Illustrates 6 · Tracks 4 View all ↓Collapse ↑

Regional expansion has moved from isolated demonstrations toward coordinated development of platforms, standards, project schools, and learning centers. Shenzhen’s Futian District reports progress that includes 20 project schools, 78 AI learning centers, and application standards. These scale indicators still need to be connected to teachers’ active use, classroom tasks, student work, urban–rural differences, and conditions for replication. [46][47][48] plus 23

26 sources for this claim Supports 3 · Illustrates 11 · Tracks 12 View all ↓Collapse ↑

Classroom evidence is moving from course counts toward tasks, work products, and process records. A Beijing youth practice program involved 201 schools and 963 projects, a Puning competition recorded 405 participants, and a national challenge reported 170,000 registrations. These figures show participation routes, while routine process assessment, students' independent contribution, and broad learning outcomes still need separate verification. [72][46][73] plus 6

9 sources for this claim Supports 2 · Illustrates 4 · Tracks 3 View all ↓Collapse ↑

WHAT WE STILL DON'T KNOW

Limits of this answer

Public evidence is still insufficient to support the following judgments. Future updates will prioritize original material that can narrow these gaps.

01
Most available public information concerns policies, class hours, platforms, and training; comparable student learning outcomes across regions remain limited.
02
Counts of covered schools, registered teachers and students, and training attendances may use different statistical definitions and cannot be substituted for one another.
03
Jiangsu’s reported 9.03 million figure is the cumulative number of summer livestream views and cannot be converted into a count of unique teachers or a training completion rate.

Weekly issues referencing this topic

How Do Five-Year School AI Projects Add Up?No. 003 · 2026-08-31
Who Teaches Students to Use AI After School?No. 002 · 2026-08-24
Which AI Outcomes Are Worth a School’s Investment?No. 001 · 2026-08-17