What China is doing in AI literacy

🏫 Schools

• Progression by stage: in the early grades, notice and experiment; in the upper elementary and middle school grades, understand and apply; in high school, create projects.
• Maybe standalone subject or enroll in Computing, Science, Technology, and Practical Activities.
• Free national platform with sections on “learning AI” and “using AI,” also open to parents. The 2026 plan directs rural schools to use this platform.
• 184 model schools test and spread models.

🎓 Universities and work

• AI as a foundational subject at all universities, with materials for each subject area.
• Teachers: national AI literacy standard, large-scale assessment, AI in teacher education programs, and in the teacher certification exam.
• New official professions, such as “generative AI application operator” (2024).
• Vocational education revised with AI, microcredentials, and courses through the open university.

👵 Population

• National AI literacy initiative since 2026, including older adults and safe use.
• Servers: ~300,000 in Zhejiang. Party schools teach people to use large models to draft, analyze data, and make decisions.
• Older adults: DeepSeek classes at universities for older adults. In Guangzhou, 60 spots filled in 1 minute.
• Small businesses: 100 pilot cities for digitalization and computing and model vouchers to make AI more affordable.

✅ What’s real

• Scale and coordination: in 18 months, a notice, two guides, a five-agency plan, and provincial targets with required hours were issued.
• Clear usage rules, by age, with specific prohibitions.
• Think about those who teach: AI literacy standard for teachers and AI on the certification exam.
• Reaches overlooked audiences: older adults, public servants, and small businesses.
• Free national platform for schools with limited resources.

⚠️ What’s concerning

• Narrow mandate: Eight hours a year is little. The phrase “mandatory nationwide from age 6,” common on foreign websites, does not appear in official texts. National policy sets guidelines, while each province sets the specific requirements.
• Inequality: rural schools are short about 8.5 million computers. In the 2025 college entrance exam, 85% of Shanghai students entered a four-year university, compared with 32% in Anhui.
• Companies inside schools: iFlytek charged ¥1.74 million (about US$254,000) to a single school for its “smart classroom” package.
• Surveillance: students’ brain-wave patterns (2019, later suspended after protests). In 2025–26, visitors reported cameras in classrooms and hallways and students wearing devices in Shanghai.
• Ideology: the 2026 plan calls for a “large-scale political and ideological education model” and digital profiles for each student.
• No measurement: there is still no independent study of what students have learned.

1. Progression by level

Recognize, understand, create. The same progression can help organize an AI learning path for beginners, starting with an introduction and ending with their own projects.

2. A simple use policy

A list of approved tools, with rules not to copy and paste sensitive data and to verify information with another source. It fits on one page and works for schools, families, and small businesses.

3. Train teachers first

China defined what teachers need to know before expecting students to learn it. Well-trained multipliers deliver more than scattered classes.

4. People 40+ and 60+ want to learn

AI classes for older adults sold out in one minute. The demand is there, and there’s often a lack of offerings designed for this audience.

5. Public administration

AI course for civil servants and city governments, modeled on Zhejiang: writing, analyzing data, and making decisions with AI support.

6. What not to copy

Surveillance, dependence on a single vendor, and literacy without measuring results. Open, free, neutral content is the counterpoint.