Introduction

In 2025, the State Council issued the “Opinions on Deepening the Implementation of the ‘Artificial Intelligence+’ Action,” which outlined the overall requirements, development goals, and key directions for promoting the deep integration of artificial intelligence (AI) across various sectors of the economy and society. Following this, the “Beijing AI Education Work Plan for Primary and Secondary Schools (2025-2027)” was introduced, establishing AI as a core engine for driving high-quality development in basic education.

Education Minister Huai Jinpeng noted that AI brings new opportunities for education while also posing challenges in responding to technological revolutions and industrial transformations. As a foundational stage for talent cultivation, primary education must break through the limitations of traditional teaching models and deeply integrate AI technology throughout the educational process, transitioning from a “one-size-fits-all” teaching approach to one that empowers differentiated growth.

The Fengtai Experimental Primary School of the Beijing Academy of Educational Sciences, as a national AI education experimental school, has adhered to the educational philosophy of “differentiated growth” since its establishment in 2012. Under the guidance of party building, the school has developed an ecological education system called the “Fertile Soil Curriculum”. After 13 years of exploration, the school has formed an AI education practice system that encompasses curriculum development, teaching practice, and evaluation reform, meeting the diverse growth needs of students.

Multifaceted Practical Pathways: Systematic Iteration of Curriculum, Teaching, and Evaluation

Firstly, the school has restructured its curriculum system, achieving a three-level integration of AI in course design.

Basic courses utilize AI-enabled differentiated teaching methods, deeply embedding AI elements into subject instruction. Tasks are designed based on students’ cognitive differences: In the third grade information technology class, Bingo robots and AI general knowledge courses are introduced to awaken technical awareness through programming and hands-on practice. In the lesson “Brushing Li”, students engage in a task called “I Represent Cao Xiaosan” using the “Feng Jicai AI Video”. Most students grasp the emotional trajectory of Cao Xiaosan. In the lesson “Ancient Versus Modern”, AI dynamically presents the evolution of Chinese characters and assists in reading training. In mathematics, an intelligent platform simulates a basketball game, allowing students to adjust parameters and observe changes in shooting accuracy, thereby visualizing statistical concepts. For homework, the AI assigns practice exercises based on student performance in the “Ancient Versus Modern” lesson, providing precise support for differentiated growth.

The school offers extension courses in intelligent robotics and programming, including a 3D modeling creative workshop and BT robot programming. Project-based learning activities such as “I Participate in Playground Renovation” and “Crazy Roller Coaster” allow students to use AIGC (AI-generated content technology) tools to transform ideas into visual outcomes. In collaboration with the Hong Kong Lok Sin Tong Leung Kau Kui School, the dual-teacher classroom “Crazy Roller Coaster” utilizes real-time video connections, enabling students to upload design diagrams and test videos through corresponding software, sharing and evaluating in real-time, and continuing to interact and optimize their work outside of class. This implements interdisciplinary AI practice projects.

The school standardizes extracurricular clubs for sports, technology, and arts, developing personalized courses in AI-assisted creation and intelligent programming to support students’ interests. The Art Academy conducts the “AI Design of Jing Embroidery Patterns” project, where students optimize creative plans using AI tools. The Science and Innovation Academy offers a robotics club and maker lab, where students complete exploratory projects using programming and sensors. In the Labor Academy’s “Campus Vegetable Garden” project, students use AI tools to collect data on plant growth and analyze the impact of environmental factors.

Secondly, the teaching model has been innovated, with AI applications implemented in multiple scenarios.

Each subject conducts targeted integration of AI practical applications: In information technology, the “Huan Xiao Zhi” intelligent agent explains professional knowledge, aiding STEAM (Science, Technology, Engineering, Arts, Mathematics) problem-solving; in art, a digital person named “Pan Qiqi” is introduced to assist in teaching and evaluating artwork; in music, AI is used to compose songs and generate musical animations; in English, the Youdao Dictionary Pen shortens word lookup time and provides standard pronunciation.

During intelligent classroom observation and diagnosis, the Zhongqing AI classroom observation diagnosis system is introduced, utilizing intelligent cameras and voice recognition technology to record classroom behaviors and emotional states throughout the lesson, generating multidimensional analysis reports. Teachers can optimize teaching strategies based on these reports. For example, in mathematics, the system automatically identifies students’ conceptual misunderstandings and directs visual tools and tiered training questions, achieving precise intervention.

In differentiated teaching, tools like the “Huan Xiao Zhi” intelligent agent are used to construct a closed loop of “differentiation recognition - dynamic intervention - feedback optimization”. Intervention courses are designed for children with ADHD, combining sand table, sensory integration training, and AI attention training. For students with learning difficulties, basic learning resources and gamified tasks are provided, while students with advanced capabilities are offered challenging projects and innovative guidance, achieving a tailored approach for each individual.

Lastly, the evaluation system has been reformed, constructing a comprehensive digital evaluation framework.

A three-dimensional evaluation system is established, focusing on “foundation building - expansion building - personalized building”, incorporating tools for AI essay grading and intelligent physical fitness monitoring. The foundational evaluation emphasizes mastery of core knowledge through unit tests and process data tracking; the expansion evaluation focuses on interdisciplinary practical abilities, using project reports and peer evaluations; the personalized evaluation highlights the development of interests and strengths, establishing awards like “Creative Star” and “Thinking Breakthrough Award” to recognize personalized growth highlights.

Dynamic evaluation and feedback are implemented, utilizing AI technology to achieve full-process digitalization of evaluation. In class, learning data is collected through gesture feedback and instant response devices; after class, the “Class Optimization Master” records students’ dynamic performance, generating visual growth maps; at the end of the term, the SOLO (Structure of Observed Learning Outcomes) taxonomy and AI analysis tools are used for multidimensional evaluations across five subject areas and nine core competencies, replacing the traditional “score-only” model.

A diverse evaluation community is constructed, involving teachers, students, parents, and AI. Teachers conduct targeted evaluations based on AI reports; students engage in self-reflection through growth portfolios; parents provide feedback on students’ home learning situations through school-home platforms; AI tools offer objective data support, forming a comprehensive evaluation loop.

Addressing Core Issues: Enhancing Competencies and Optimizing Ecology

Through continuous exploration, four core issues have been effectively addressed. 70% of the school’s courses have achieved interdisciplinary design, and most teachers can independently develop interdisciplinary projects. These courses have increased student engagement and improved their comprehensive abilities, making the integration of five educational aspects more effective. Teachers design foundational and challenging questions in class, ensuring that students at different levels benefit from learning. Surveys indicate a significant decrease in the proportion of students spending over two hours on homework. Most teachers can adopt differentiated task designs based on each student’s situation, solving the “one-size-fits-all” problem in class instruction. 89% of teachers are proficient in using intelligent tools, with most classes achieving data-driven precise interventions, shifting digital empowerment from a superficial to a deeper level. The evaluation system has fundamentally changed from being score-centric, with students now paying more attention to moral character and practical non-academic evaluations.

Students’ competencies have been comprehensively enhanced. With the empowerment of digital tools, the accuracy of differentiation recognition has improved, and teachers provide personalized guidance based on the “one student, one strategy” reports, leading to increased classroom focus and proactive participation among students with learning difficulties. All students can participate in interdisciplinary courses represented by “Jing Embroidery Imagination” and “Subway Research”. Students express satisfaction with personalized courses, and 90% of middle school teachers report that our students’ “independent inquiry and collaboration abilities exceed those of their peers”.

The educational ecology continues to optimize. From the perspective of model innovation, the implementation paths of “Fertile Soil Curriculum” and “Digital Intelligence Empowerment” have been constructed, achieving deep integration of AI with the five educational aspects and differentiated growth. A closed loop of “data collection - intelligent analysis - precise intervention” has been established, developing seven types of digital intelligence courses and intelligent agents like “Huan Xiao Zhi”, improving the precision of differentiation recognition. A comprehensive digital evaluation model has been established, forming a complete chain of “process diagnosis - personalized feedback - developmental incentives”, effectively addressing the issue of score-centric evaluations.

AI provides a new empowering path for differentiated growth in primary school students. The school’s 13 years of practice demonstrate that through the comprehensive integration of curriculum system reconstruction, teaching model innovation, and evaluation system reform, many challenges in traditional primary education can be effectively addressed. In the future, the school will continue to adhere to the principle of “student-centered, technology-enabled” in AI education, exploring and innovating continuously to enable every student to achieve differentiated and high-quality growth in an AI-empowered educational ecology.