AI is reshaping education with unprecedented depth and breadth. Yet the stronger the technology becomes, the more firmly education should focus on the all-round development of people, especially the human capabilities that technology cannot easily reach.

For a generation of digital natives, how should education systems respond to the opportunities and challenges brought by AI? At the recent 2026 World Digital Education Conference, representatives from academia, schools, and local education authorities discussed the theme of cultivating thinking abilities beyond artificial intelligence.

Defining the Core Qualities of Future Talent

When students can conveniently use AI tools to complete assignments and obtain answers, what abilities do they truly need that AI cannot replace? Participants argued that thinking beyond AI is not an abstract idea, but a system of specific, teachable higher-order human qualities.

Li Yongzhi, president of the National Institute of Education Sciences, said knowledge and ability are not opposites. Future education will still need knowledge learning as a foundation, but should devote more energy to developing abilities beyond knowledge acquisition.

Li said the key to surpassing AI is intrinsic motivation. The foundation for building and applying abilities is inner drive: as carbon-based organisms, people naturally develop the drive to eat and survive, while machines have no such drive and depend entirely on preloaded rules and instructions. This motivation, rooted in interest, confidence, meaning, and social-emotional needs, is the fundamental force that enables people to navigate the future and move beyond machines.

Critical thinking and complex problem-solving were among the most frequently discussed themes. Andreas Schleicher, director for education and skills at the OECD, cited a thought-provoking case from Turkey: after students used AI tools to learn mathematics, their grades improved, but their mathematical thinking declined in later tests.

Schleicher warned that AI tools raised test scores but did not improve learning ability. The case reveals a risk: if technology is used improperly, it may weaken students’ capacity for deep thinking and independent exploration.

In knowledge storage, rapid retrieval, and rule execution, AI has already far surpassed humans. Education therefore needs to re-anchor itself in distinctively human value. Li Yongzhi argued that education should pay more attention to higher-order thinking, reflection, empathy, and the ability to make ethical judgments and choices.

Huang Changqin, director of the Zhejiang Key Laboratory of Intelligent Education Technology and Application, further explained that warm education must include positive value guidance and a sense of humanistic belonging. Students need spiritual coordinates, emotional connections, and cultural identity, which are precisely the areas cold algorithms struggle to reach.

Practice has shown that students need to learn not how to depend on AI, but how to collaborate with it. Mastering and using AI well is an essential survival skill in the digital age. Zhu Xinyu, deputy director of the Institute of Educational Statistics and Analysis at the National Institute of Education Sciences, said that in the AI era, education must cultivate not only students’ ability to use technology, but also wisdom beyond technology, including how to coexist with AI, ensure technology is used for good, and bring out the unique value of people.

Systematic Approaches to the Problem

At the 2026 World Digital Education Conference, the Global Digital Education Development Index 2026 was officially released. It evaluates and tracks education development in 82 countries in the AI era, and for the first time included the cultivation of thinking abilities beyond AI as one of its research dimensions.

The data show that 78% of countries believe education in the AI era should emphasize students’ higher-order thinking. In terms of the components of student thinking ability, more than 50% of countries regard AI application ability, critical thinking, ethical judgment and decision-making, creative thinking, problem-solving ability, and social-emotional ability as key student capabilities in the AI era.

In analyzing countries’ digital education development paths through the index, Zhu Xinyu said China’s digital education is characterized by systematic advancement. It places relatively strong emphasis on national-level top-level design, and uses infrastructure development to support more balanced education, including the construction of the world’s largest digital education resource center and platform.

Zhejiang’s actions are representative. Zhu Hongping, president of the Zhejiang Academy of Educational Sciences, said Zhejiang treats AI as a key variable in education modernization reform and has built a general AI education system covering all stages of schooling.

In basic education, Zhejiang has iteratively developed platforms for science, technology, and AI learning. In higher education, it has achieved full coverage of general AI courses for first-year university students. On this basis, Zhejiang is working to build technology-enabled future classrooms. It has also released a teacher AI literacy framework and carried out dedicated training to improve teachers’ leading role as designers and collaborators. Zhu Hongping said that in this era, the leading role of teachers has not been overturned, but strengthened and enriched.

Some schools reported that the teaching paradigm is shifting from knowledge transmission to thinking development, and that profound changes are already taking place in classrooms.

Li Yongzhi shared one example. An eighth-grade student at a middle school in Beijing used agents to complete one semester of a course independently and efficiently. The key was that the student did not passively accept AI-generated answers, but actively designed the learning process: asking agents to analyze the course content into logically connected knowledge points, generate personalized learning materials such as audio and video, make full use of fragmented time, and conduct targeted tests and error analysis.

This case inspired the National Institute of Education Sciences project on multi-agent collaboration based on a teaching chain of thought. Li explained that the point of the example is not to encourage students to leave the classroom, but to transform the classroom. Learning processes such as learning-status analysis, resource generation, and instructional design can be broken down and supported by specialized agents, while another specialized agent coordinates the process. This can free teachers and students from repetitive training and use the saved time to cultivate future-oriented key abilities beyond AI.

Educational evaluation is the command baton. In the AI era, how can evaluation reform make each person’s growth visible? Chen Liang, director of the Education Bureau of Gongshu District in Hangzhou, shared a regional practice of using AI to support education evaluation reform: a literacy-oriented project-based assessment model.

The model translates educational goals into observable performance in real tasks, and uses AI for data collection, analysis, and profiling. Chen said that in non-paper-and-pencil assessments for lower primary grades, for example, qualitative definition, rule setting, quantification, and profiling are used to create process-based, evidence-based evaluations of students’ comprehensive literacy. The goal is to turn evaluation from a ruler that measures results into a mirror that sees process and understands growth, ultimately driving teaching improvement and better governance so that education becomes fairer, higher-quality, and warmer.

Building Educational Rationality in the AI Era

While embracing technology and innovation, participants also clearly identified current problems and potential risks. Experts proposed constructive responses and called for a healthier and more controllable ecosystem for AI applications in education, promoting human-machine coexistence.

Some experts raised the idea of suspended capability. AI provides unprecedented opportunities for everyone to enhance their abilities, but if individuals or groups fail to master and use these capabilities effectively, the opportunities become meaningless. If this suspension is unevenly distributed because of differences in socioeconomic status, region, or digital literacy, it will create a new intelligence divide.

Huang Changqin said the inclusiveness of public digital infrastructure must be strengthened. His proposed response is to make public facilities broadly accessible and build a fair, adaptive digital foundation. Zhejiang’s provincial data base, Education Cube, and its AI Can Learn platform are attempts to coordinate computing power and resources across the province and narrow gaps between regions and schools.

Superficial cognition and intellectual laziness are among the most concerning challenges. Excessive dependence on AI may lead students into a fluency trap: they become satisfied with smooth AI-generated answers, reduce their own thinking effort, and form a habit of cognitive outsourcing. Schleicher cited a U.S. case showing that among students who used AI to write essays, 80% could not remember what they had written. Survey data from the National Institute of Education Sciences also show that 85.6% of primary and secondary school students have used AI to complete homework.

How can education uphold the essence of learning as effortful engagement? Experts said instructional design should break the fluency trap and encourage deep cognitive participation. One method proposed by Dragan Gasevic, a professor at Monash University in Australia, is worth considering: design tasks that require students to question, verify, or generate questions about AI-produced information; organize group discussions comparing different AI outputs or human-AI viewpoints; and cultivate critical thinking through comparison and debate.

The deep integration of AI and education is unstoppable. The key to this transformation is not the brilliance of the technology itself, but how we use technology to awaken, nourish, and protect the precious traits that make humans human: curiosity, empathy, critical thinking, creativity, and unending intrinsic motivation.

As one participant summarized, the road ahead is full of challenges. But the educational rationality needed in the AI era is one in which AI becomes scaffolding for thought, not a crutch that replaces thinking. The new learning paradigm should be an autonomous journey with technology at one’s side and thinking in one’s heart. The development path of AI plus education should turn the cultivation of thinking abilities beyond AI from a shared concept into vivid practice and institutional support rooted in China.