Youth learns to transform future with AI

At the top end, the strongest performers in informatics Olympiads may gain admission to elite universities such as Tsinghua University and Peking University without taking the national college entrance exam (known as Gaokao).

When the generative AI boom came, children who had spent years learning to code or build robots suddenly felt that AI was close at hand. It was something built by people not much older than they were, rather than another technology imported from distant Silicon Valley.

Liang Wenfeng of DeepSeek and Yang Zhilin of Moonshot AI, along with Wang Xingxing of Unitree Robotics, have become new role models for technology.

During the 2026 summer vacation, AI-themed study tours are gaining ground on the English-language and military-themed camps that once dominated the market, with bookings up 370 percent, according to online travel platform Ctrip.

Preparing Children

The story does not end with after-school clubs and competitions, though. An action plan released in April calls for AI courses to be introduced across all levels of education, from primary schools to universities.

The aim is to prepare children for the society they will grow up in, says Xiong Zhang, a computer science professor at Beihang University who heads an expert panel responsible for developing information technology curriculum standards for compulsory education.

"If they are to adapt to the future, they need to start learning about AI, understanding it and learning how to work with it from an early age," he says.

This idea is not unique to China. Fei-Fei Li, a Stanford professor often dubbed the "Godmother of AI", has stressed the importance of public AI education. She has likened the skill of asking AI the right questions to Socrates' method of seeking truth through questioning, arguing that the quality of the questions can shape the quality of the answers.

Some parts of China have already moved ahead. Schools in Beijing, for example, began offering general AI education from primary through senior high in the fall semester of 2025.

As in many countries in Europe and North America, however, AI is evolving so quickly that schools vary in teacher expertise, curriculum development and access to computing infrastructure and other hardware.

China is therefore encouraging schools to take an interdisciplinary approach — one that makes particular sense given that AI draws on fields including mathematics, linguistics, psychology and biology.

"This is both a practical way to get AI literacy education off the ground and an important foundation for deepening it in the years ahead," says Guan Bo, a researcher at a center studying the digital transformation of education in Guangxi Zhuang autonomous region.

The approach evolves as students grow older. For younger children, a national guide released in 2025 calls for games and hands-on activities. By junior high school, students are encouraged to connect AI with their regular subjects — building, for example, a small autonomous vehicle in physics and learning how sensors and controllers turn perception into action.

By senior high, AI education becomes more research-oriented, with students tackling projects that require them to identify problems, design and test solutions, and present their findings.

That approach is already visible in the competitions. At the AI-agent contest in Wuzhen, Shi Jiaxing, an 11th-grader from Sichuan province, presented a system designed to spot signs of emotional distress in young people and provide online mental health support.

The system uses multiple AI agents and safety guardrails to reduce the risk of mistakes. A lightweight training method allows it to run on basic devices, potentially making AI-powered mental health support more accessible to adolescents in rural areas.

IResearch, a research and consulting firm focused on China's new economy, estimated in February that spending on informationization and the digital transformation of education in China would reach 551.5 billion yuan ($81.9 billion) in 2025.

Some experts, however, warn against another extreme: allowing students to outsource memorization, calculation, reading, writing and basic reasoning to AI while retaining little more than "creativity" and "judgment".

Xiao Yanghua, a professor at Fudan University's College of Computer Science and Artificial Intelligence, says this is another pitfall that education reform must guard against.

"Without a solid foundation in core cognition, it is difficult to develop reliable higher-order thinking," Xiao says. "Students without a well-structured body of knowledge may not be able to recognize where AI has gone wrong."

Reframing science CLASS

In a document released last year on the use of generative AI in primary and secondary education, China outlined typical use cases and warned against students simply copying AI-generated content in their schoolwork or relying too heavily on generative AI for assignments that call for creative or personal expression.

In the meantime, China is not framing AI as something that should crowd out everything else in a child's education.

In its education planning through 2030, the country continues to prioritize character development and students' well-being, including a health-first approach to schooling. AI education is seen as part of a broader effort to develop students' abilities and skills, rather than as an end in itself.

In fact, AI's growing role in knowledge production is prompting a broader rethink of how science should be taught in China.

This autumn, schools nationwide will launch an initiative called "learning through doing", which experts say could reshape science classes in China's compulsory education system.

Students in grades four through nine will be required to complete at least one inquiry-based, hands-on project each semester, with each project spanning at least four class periods.

"Trial and error is at the heart of science," says Fang Yu, a chemist and an academician at the Chinese Academy of Sciences.

In the age of AI, when information is increasingly easy to access and standard answers can be generated in an instant, science education, he says, can no longer center on memorizing facts and following prescribed procedures. It must return to the fundamentals: asking questions, learning from failure and exploring what is not yet known.

Consider a familiar classroom experiment: producing carbon dioxide. Traditionally, Chinese students might follow a prescribed sequence of steps toward a predetermined conclusion.

Under the new approach, they might test different combinations of acid concentrations and carbonate amounts, observe how gas output changes, troubleshoot leaks in the apparatus, and work out why their results differ.

"The shift goes beyond teaching methods. It points to a fundamental change in what education is meant to cultivate," Fang says.

XINHUA