Chongqing's Nan'an District Unveils AI Education Blueprint: One Study Companion per Student by 2030
Nan'an District of Chongqing has published the Ten Measures for Deepening AI + Education in Primary and Secondary Schools, a district-level blueprint that sets five 2030 goals: one high-quality regional education dataset, one cross-disciplinary teaching workforce, a group of innovation-focused schools, at least one signature AI application scenario in every school, and one AI study companion for every student.
The measures were made public around the 2026 TongAI Conference on General Artificial Intelligence held in Chongqing on October 22. Local outlets including Chongqing Daily and Xinhua's Chongqing channel reported the release, describing it as the first district-level AI education programme of its kind in the municipality.
On infrastructure, the district plans future learning centres and smart teaching equipment in all schools, and will pilot an education computing voucher scheme that pools computing demand at district level and allocates it to schools as vouchers, so that even under-resourced schools are not priced out of AI tools. The plan also targets 100% of students completing 100% of their experiments.
Data will be aggregated through Digital Chongqing's integrated intelligent public data platform, known as IRS, to build a regional multimodal corpus and develop an education-specific large model tuned to local teaching conditions rather than relying on general-purpose models.
For teachers, the measures set a staffing ratio of one full-time or part-time AI teacher for every six classes, and call for five municipal-level and 100 district-level seed teachers who will in turn reach more than 5,000 teachers, with training covering the entire workforce.
Students will follow mandatory AI literacy hours: at least 10 class periods in grades 1 and 2, 25 in grades 3 to 6, and 30 in junior and senior high school, organised into a three-tier pathway of awareness, application and innovation. After-school services must include at least one AI interest class per week, and every October will be designated a primary and secondary AI month with an annual innovation competition.
The district also plans 20 district-level AI feature schools and aims for three to five municipal base schools, a round-the-clock reading assistant bot, an AI plus brain science learning capability laboratory built with the Chongqing Brain and Intelligence Science Centre, a pilot for scientist digital avatars, and at least one AI study trip per school each semester.
To address concerns about AI-assisted cheating, the education-specific model will, when answering questions, give only approaches, hints and follow-up questions rather than direct answers. Three hard lines are set: AI may not be used to complete entire homework assignments, may not generate harmful content, and may not be used without supervision.
An official from the district education authority acknowledged the difficulty, saying implementation is genuinely hard because it tests platforms, computing power and teacher capability. The working group kept the companion goal anyway, arguing that it is the most direct way AI can help students, since every child gets an on-demand, always-available learning partner, and that this is the core meaning of AI plus education.
On equity, the official said the district is not building a few showcase schools, and will apply unified class-hour standards, terminals and cloud resources across weak and strong schools alike. Generative AI tools must pass a safety assessment before entering campuses, and biometric data such as faces and voices will be deleted immediately after class and barred from commercial use.
The measures also create a Chengdu-Chongqing AI education partner-school network that links Beijing, Shanghai, Hangzhou and Shenzhen, and reaches outward through sister-school ties. The stated aim is a demonstrative zone that is leading in western China and recognisable nationwide.
The measures were made public around the 2026 TongAI Conference on General Artificial Intelligence held in Chongqing on October 22. Local outlets including Chongqing Daily and Xinhua's Chongqing channel reported the release, describing it as the first district-level AI education programme of its kind in the municipality.
On infrastructure, the district plans future learning centres and smart teaching equipment in all schools, and will pilot an education computing voucher scheme that pools computing demand at district level and allocates it to schools as vouchers, so that even under-resourced schools are not priced out of AI tools. The plan also targets 100% of students completing 100% of their experiments.
Data will be aggregated through Digital Chongqing's integrated intelligent public data platform, known as IRS, to build a regional multimodal corpus and develop an education-specific large model tuned to local teaching conditions rather than relying on general-purpose models.
For teachers, the measures set a staffing ratio of one full-time or part-time AI teacher for every six classes, and call for five municipal-level and 100 district-level seed teachers who will in turn reach more than 5,000 teachers, with training covering the entire workforce.
Students will follow mandatory AI literacy hours: at least 10 class periods in grades 1 and 2, 25 in grades 3 to 6, and 30 in junior and senior high school, organised into a three-tier pathway of awareness, application and innovation. After-school services must include at least one AI interest class per week, and every October will be designated a primary and secondary AI month with an annual innovation competition.
The district also plans 20 district-level AI feature schools and aims for three to five municipal base schools, a round-the-clock reading assistant bot, an AI plus brain science learning capability laboratory built with the Chongqing Brain and Intelligence Science Centre, a pilot for scientist digital avatars, and at least one AI study trip per school each semester.
To address concerns about AI-assisted cheating, the education-specific model will, when answering questions, give only approaches, hints and follow-up questions rather than direct answers. Three hard lines are set: AI may not be used to complete entire homework assignments, may not generate harmful content, and may not be used without supervision.
An official from the district education authority acknowledged the difficulty, saying implementation is genuinely hard because it tests platforms, computing power and teacher capability. The working group kept the companion goal anyway, arguing that it is the most direct way AI can help students, since every child gets an on-demand, always-available learning partner, and that this is the core meaning of AI plus education.
On equity, the official said the district is not building a few showcase schools, and will apply unified class-hour standards, terminals and cloud resources across weak and strong schools alike. Generative AI tools must pass a safety assessment before entering campuses, and biometric data such as faces and voices will be deleted immediately after class and barred from commercial use.
The measures also create a Chengdu-Chongqing AI education partner-school network that links Beijing, Shanghai, Hangzhou and Shenzhen, and reaches outward through sister-school ties. The stated aim is a demonstrative zone that is leading in western China and recognisable nationwide.
JMDedu is a leading B2B EdTech Media in China that informs and connects business professionals through our reports, events, and information services. Our mission is to drive the advancement of education. Our network spans the education industry in China as well as globally, including entrepreneurs, investors, startups, educators, business decision-makers, and policy decision-makers.
Copyright © 2014-2026 en.jmdedu.com All Rights Reserved