ChinaTech - China's Guidelines for Generative AI in Education
... developing an AI-native and tech-literate generation
This segment by Shobhankita Reddy is your go-to newsletter for updates and perspectives on China’s tech ecosystem. This edition studies the latest guidelines by the Ministry of Education for the use of Generative AI in schools.
Last week, the Ministry of Education in China issued two guidelines on the use of Generative AI in schools. The documents are available in full here and here.
Early this year, it was reported that AI education for primary and secondary school students would be made mandatory starting September 2025 and for students as young as six. At the time, the programme entailed at least eight hours of age-specific AI education for students of all levels annually and was part of a national plan to support the modernization of education and develop a tech-literate generation. A white paper detailing the specifics of such a plan was said to be in the works. This is now out and is very interesting.
First, a few key pointers. Unlike the UNESCO recommendation of an age limit of 13, China envisages an earlier and more foundational integration of Gen AI for its students. Secondly, both documents have been referred to as "guidelines" and don't seem to be legally binding regulations. Their value appears to be in signalling intent, conveying the Ministry's thinking around best practices of incorporating AI into education, encouraging bottom-up experimentation and laying the groundwork for the future. Also, China's guidelines go beyond merely regulating the use of the technology and placing constraints on different stakeholders by way of requirements for responsible AI or data privacy - although there is some of that. The guidelines are more a call to action and are important for their detailed thinking on the deployment of Gen AI - the many scenarios it can be used for by teachers, parents, school administrators and students as well as the learning outcomes and different age-appropriate focus areas it can further. These have been divided into four pillars: technical knowledge, practical skills, critical thinking and ethical awareness.
The technical knowledge envisaged moves from cultivating interest in Gen AI and perceiving the value of Gen AI in basic technologies like image classification and speech recognition at the primary school level, to understanding technical logic such as supervised learning, algorithm selection and data features at the middle school level and finally to understanding the strategic societal impact and practical applications in national strategies (such as smart city missions and national defence security) at the high school level. A similar progression in practical skills to be cultivated in students follows across the three tiers of a student's education - from basic application capabilities such as visual programming at the primary level to constructing simple algorithms with optimized performance at the high school level.
Similarly, the critical thinking pillar starts with cultivating basic logical thinking and differentiating between AI and human behaviour at the primary level. It moves on to developing "engineering thinking" at the middle school level. The goal at the high school level is to develop "interdisciplinary system thinking for use in innovative projects". These are aimed to be balanced by focusing on privacy and digital identity protection at the primary level, teaching the risks of misinformation and disinformation at the middle school level, and aligning technological innovation with societal and ethical risks in more complex scenarios at the high school level.
The implementation strategies put forward for this to happen are summarized as follows.
Classroom Education - Improve the curriculum by incorporating AI education in Information Technology, science, and other courses. This should be tiered depending on the needs of students across ages and different cognitive levels. Schools are encouraged to explore innovative teaching methods for lectures, projects, case analyses and enhancing the fun and effectiveness of teaching.
Organization and Implementation - This includes combining AI education in campus activities such as cultural activities,science and technology festivals, technology challenges, innovation project exhibitions and evaluations and after-school services) and off-campus activities, improving resource sharing and collaboration between schools and involving family participation, technical personnel and industry experts in the teaching process.
Teaching Evaluation - Schools must develop an evaluation index and mechanism across the four dimensions of "knowledge - skills - thinking and values" combining the inputs of teachers, students and parents, which would then be used to improve teaching, share measures that work with other schools, and establish an innovative-incentive mechanism to promote overall AI education.
The guideline also envisages "actively introducing qualified professionals from universities, research institutes, and high-tech enterprises to serve as part-time AI education teachers", increased infrastructure support - "set up AI education bases", "promote the opening of AI laboratories, exhibition halls and other venues of universities, research institutes and high-tech enterprises to primary and secondary schools", and strengthening urban-rural coordinated development to ensure fair and inclusive development.
The need for AI products and services in these schools is worth mentioning. Social enterprises have been called upon to help integrate education resources with industry:
Encourage leading AI companies and educational technology companies to develop highly adaptable and scientific teaching tools and course products based on advanced technology and educational experience, based on the cognitive characteristics of primary and secondary school students and AI course requirements, and accelerate the construction of high-quality and professional AI education products and services.
The second document lays out the different applications and usecases in which AI can be deployed in education, and it includes several safeguard measures to be aware of. It is highly detailed and contains many examples.
Promoting student growth -
Focusing on the diverse learning and growth needs of students at different educational stages and of different types, generative artificial intelligence is applied to specific scenarios to provide personalized support and guidance to promote the all-round development and healthy growth of students.
This includes personalized learning, interactive exploration, improving students' mental health, and ensuring equal learning opportunities for students with special needs. The examples in each of these subsections are particularly instructive in the many applications and use cases Gen AI can employ. For instance,
For example, students can rely on the AI reading companion system to analyze reading trajectories in real time, obtain personalized suggestions and interactive reading questions; young students can use dynamic picture book generation tools to communicate with virtual characters that incorporate elements of traditional culture and red culture to promote interactive narratives; call on the multimodal audio book system to experience the multi-dialect and emotional voice interpretation of classic texts; generate visual knowledge graphs based on academic literature analysis functions, and deepen the cognitive system in combination with cross-cultural background analysis.
Assisting teachers with teaching - tailoring teaching methodology to the needs of the course and the students' conditions across various formats of text, audio and video, enabling classroom interactions by leveraging Augmented Reality (AR) and Virtual Reality (VR), after-school evaluations such as plagiarism detection, diagnostic feedback etc, and to recommend further homework and tutoring plans.
For school departments, GenAI use cases span a wide range—from reducing the burden on administrative tasks such as preparing meeting minutes, work reports, multi-language translations, etc., to matching capacity when there is a shortage of teaching staff.
For example, paper documents can be digitized and converted into editable electronic texts through optical character recognition (OCR) technology, and then generative AI can be used for text analysis and information extraction; key information such as student grades, curriculum settings, and teacher evaluations can be extracted from archives to generate structured data to provide decision support for education management; teaching materials and research results in historical archives can be analyzed to extract valuable teaching methods and research ideas to assist teachers in improving teaching practices and conducting educational research.
As for safeguard measures, Education administrative departments must -
adhere to the basic principles of "adapting measures to local conditions, implementing policies in a classified manner, and standardizing management". Fully consider objective factors such as regional development imbalances, urban-rural resource differences, and local characteristics and diversity, and scientifically formulate application promotion paths in this jurisdiction. Adhere to the regulatory concept of "inclusiveness, prudence, grading and classification", establish and improve the regulatory system and "clarify the list of generative artificial intelligence tools that can be used in schools".
Primary and Secondary schools —
Avoid adopting a simplistic management model of "one size fits all." Be vigilant against excessive reliance on AI tools, and strengthen data security and privacy protection governance. Take into account the dual goals of improving the practical skills of teachers and students and meeting emotional needs, and ensure the integration of technological empowerment and humanistic care.
Teachers -
Teachers shall not use generative artificial intelligence as an alternative teaching subject and are prohibited from directly using AI to answer students' questions or provide consultation; they should avoid directly using AI to generate content to evaluate students; it is strictly forbidden to input sensitive data such as personal information and test questions into AI tools to prevent data leakage and privacy infringement; it is not allowed to use AI to copy and disseminate other people's works without authorization to avoid copyright infringement; avoid over-reliance on AI plagiarism tools.
Students -
Students should avoid simply copying the content generated by generative artificial intelligence tools in their homework; avoid using generative artificial intelligence to take exams and tests, and not use generative artificial intelligence to cheat; avoid easily abusing generative artificial intelligence in learning tasks that show creativity or personalized expression, and lose personal thinking and opinions; avoid rashly using generative artificial intelligence to obtain information before consulting high-quality textbooks or authoritative materials; avoid entering personal information into generative artificial intelligence tools to leak data and privacy; avoid using generative artificial intelligence to copy or disseminate works without authorization, infringing copyright.
The guideline has similar supportive and supervisory roles for parents and other entities such as social enterprises and industry.
Similar to China’s approach to AI governance - iterative, incremental, experimental and distinct from anything else in the world - these two guidelines display serious, original thinking about the potential of GenAI for teaching pedagogy, student learning outcomes and the overall education system. It is likely to have come from deep deliberations and a consultative process involving multiple stakeholders.