Introduction

Artificial intelligence (AI) is reconstructing classroom ecosystems, ushering in profound changes in educational digitalization.

This year, significant policies such as the “National Education Strong Country Construction Plan (2024-2035)” and the “AI + Education Action Plan” have been introduced, highlighting the role of AI in education. At the 87th China Education Equipment Exhibition, numerous AI products related to classrooms emerged.

Amid this growing interest, key questions arise:

• Is AI in the classroom a replacement or an empowerment tool?
• How can teachers avoid being overwhelmed by massive data?
• How can deep changes in classrooms transition from knowledge transmission to nurturing competencies?

Expert Insights

Nandu N Video reporter interviewed Professor Li Yushun, a member of the first Education Informatization Teaching Guidance Committee of the Ministry of Education and director of the MOOC Research and Teaching Innovation Laboratory at Beijing Normal University. He decoded a nationwide evidence-based classroom practice:

The “Intelligent Evaluation and Diagnostic Improvement of Classroom Teaching Regional Cooperation Project” conducted by Beijing Normal University and Seewo enables nearly 100,000 teachers to use AI to reflect on their classrooms, paving a new path for large-scale, systematic transformation in education.

Accelerated Development

AI has become a key to breaking through the challenges of classroom transformation.

“The introduction of two significant policies has accelerated the process of schools and frontline teachers embracing AI,” Li Yushun stated. Since 2010, China has promoted educational digitalization as a development strategy. Previously, technology was seen primarily as an auxiliary tool, but now policies clearly indicate that AI is deeply embedded in all educational scenarios and elements, marking a shift from traditional tools to ecological empowerment, becoming an intrinsic variable in classroom transformation.

Professor Li Yushun delivered a lecture on “Evidence-Based Methods and Practices for Specialized Classrooms Based on Classroom Understanding” and released phase results.

Over the past thirty years, classroom teaching has transitioned from multimedia to internet-based to intelligent formats. However, deep challenges in classroom transformation persist: how to genuinely move from knowledge transmission to fostering students’ abilities and competencies. Li Yushun acknowledged that this requires simultaneous upgrades in teachers’ philosophies, practical professional skills, and systematic technological environments, with AI (especially generative AI) becoming the key to breaking through these challenges.

With the continuous development of AI technology, it is evolving into a “third eye” for classroom understanding, emphasizing the dual value of technological empowerment and evidence-based practices, accelerating the high-level development of classroom ecosystems.

The “AI + Education Action Plan” released in April specifically mentions: “Utilize intelligent technology to analyze classroom teaching behaviors and carry out evidence-based educational research practices.”

This work has been researched collaboratively by Beijing Normal University and Seewo for two years.

In March 2024, Beijing Normal University and Seewo officially launched the “Intelligent Evaluation and Diagnostic Improvement of Classroom Teaching Regional Cooperation Project”; in April, a cooperation agreement was signed between Seewo and Professor Li Yushun’s team, focusing on systematic exploration and practice around “evidence-based research supporting collaborative growth classrooms and excellent teacher development plans.”

After the cooperation commenced, a collaborative demonstration area was rapidly established nationwide, signing agreements with regions such as Haidian District in Beijing, Baiyun District in Guangzhou, Xiangfang District in Harbin, Tiexi District in Shenyang, Litong District in Wuzhong, and the Economic Development Zone in Hefei. This initiative supports regional teachers’ professional growth based on evidence-based research paradigms.

By the end of December 2025, Seewo’s classroom intelligent feedback system will have established 19 key application demonstration areas nationwide, covering over 5,600 schools and applied in more than 17,000 classrooms, generating over 650,000 classroom intelligent feedback reports for more than 97,000 teachers.

The “circle of friends” continues to expand. At the 87th China Education Equipment Exhibition, a licensing ceremony for the evidence-based research professional practice experimental area/school collaboration project based on classroom understanding was held, with participants including Yibin City Cuiping District, the affiliated experimental middle school of the Sichuan Provincial Education Research Institute, Chengdu No. 7 High School Yucai Campus, and the affiliated school of the Chengdu Shuangliu District Institute of Education.

Licensing ceremony for the evidence-based research professional practice experimental area/school collaboration project based on classroom understanding.

Li Yushun told Nandu N Video reporters that the explorations over the past two years align perfectly with the national strategic direction, and now is the best window period for large-scale promotion. “The introduction of policies has given us more confidence and the courage to carry this out on a large scale and as a norm.”

Human-Machine Collaboration

AI reflects, teachers make decisions, and experts guide the direction.

As AI deeply intervenes in classroom evaluation, a series of questions confront all educators: Will machines replace teachers? What roles do humans and machines play? How can we avoid teachers becoming overly reliant on data and weakening their professional judgment?

Li Yushun provided a clear answer: AI is responsible for “reflecting,” teachers maintain decision-making power, and experts calibrate the direction.

AI collects multimodal data, presents it in a structured manner, and generates diagnostic indicators, providing objective, quantifiable evidence, acting as a faithful “teaching mirror”; teachers interpret the educational significance behind the data, conduct contextualized causal analysis, choose strategies for teaching improvement, and uphold the ultimate concern for nurturing values, maintaining their professional stance; experts build frameworks, calibrate directions, and cross-validate AI’s quantitative evidence with human qualitative insights, promoting classroom understanding and preventing teachers from being misled by massive data, achieving a triangular validation of machine data, human observation, and practical experience.

“Data is not equivalent to evidence,” Li Yushun explained. Currently, many classroom reports on the market are dozens of pages long, covering hundreds of indicators, and some frontline teachers find them incomprehensible and do not know where to start. “Therefore, we particularly advocate for ‘classroom understanding’—without understanding the classroom, it is difficult for teachers’ professional growth to transition from experience to evidence-based practice.”

Based on this understanding, Li Yushun’s team systematically constructed the “Collaborative Growth Classroom Ecology Theory for Teachers and Students,” providing a theoretical reference for the collaborative restructuring of all classroom teaching elements; with “classroom understanding” as the core direction, they established a three-layer indicator analysis framework of “leading—aggregating—key,” allowing data to serve the classroom.

In practice, the “reflecting” effect is immediate.

A primary school Chinese teacher in Beijing, who had only been in the job for a year, underwent three rounds of evidence-based classroom practice through the intelligent feedback system. In the first round, AI revealed an imbalance in classroom teaching structure and an overly long introduction; in the second round, the teacher realized that the questions were fragmented and interactions were superficial; in the third round, the teacher actively optimized question design, increased varied transfer activities, and related students’ experiences and lives, allowing students to express themselves and think deeply. “Every step of improvement, from teaching structure to classroom discourse, question design, and learning activity design, was supported by data, transforming the teacher from ’teaching by experience’ to ‘modifying by evidence.’”

At Luoyang Zhongcheng Foreign Language School, the Seewo classroom intelligent feedback system covers all 70 classrooms, generating over 10,000 intelligent reports on classroom activities in one semester, leading to a collective shift in frontline teachers’ teaching methods from “experience-driven” to “data-driven evidence.”

A Chinese teacher expressed, “When the report showed that classroom lecture time accounted for 72%, I was stunned. The report made me realize that the liveliness of my classroom was merely a form of ‘false prosperity.’”

This self-awareness based on real data vividly illustrates how evidence-based research inspires teachers’ intrinsic growth motivation. As Li Yushun stated, evidence from data only highlights its significance through comparison and relevance; its value becomes deeper through focus and reflection.

From teaching structure, classroom discourse, question design to learning activity design, teachers rely on data for precise improvements, forming a positive cycle of “small cuts, deep research, and big changes.”

Regional Practices

A strategic opportunity period has begun, with common patterns applicable nationwide.

Over the past two years, regions such as Haidian District in Beijing, Baiyun District in Guangzhou, Tiexi District in Shenyang, and Litong District in Ningxia have formed distinctive and exemplary practice samples tailored to local conditions.

In Haidian District, Beijing, a three-dimensional advancement mechanism of “activity-led, school-to-school linkage, and expert support” has promoted the transformation of regional training from “collective experience discussion” to “data empirical analysis”; it has generated 625 AI classroom feedback reports for 180 teachers, covering teaching practices in seven demonstration schools, providing a replicable “Haidian Model” for the digital transformation of classroom education.

In Baiyun District, Guangzhou, leveraging a “horizontal and vertical” educational informatization layout, the research process has transitioned from traditional lesson observation to a progressive transformation of “precise lesson observation—hybrid online and offline training—AI-enabled human-machine collaborative evaluation,” achieving a full-chain digital reconstruction of pre-lesson collaborative design, in-lesson data collection, and post-lesson evidence-based diagnosis.

In Tiexi District, Shenyang, a unique “PICo” application model has emerged—quality improvement through precise analysis, using classroom observation as a starting point to analyze classrooms, teachers, and students, resulting in comprehensive quality enhancement. So far, over 19,000 reports have been generated in Tiexi District, achieving seamless AI recording and normalizing data-driven educational research, deeply integrating the classroom intelligent feedback system into teachers’ daily research processes.

Li Yushun summarized that although the samples from different regions vary, they have distilled three replicable common patterns:

1. Unified value creation in classrooms, all centered around implementing new curriculum standards to promote student competency development.
2. Consistent advancement mechanisms, employing a three-tiered linkage structure of “regional coordination—school-based deep cultivation—individual reflection,” promoting overall regional advancement, generating school-based practices, and improving teachers’ professionalism in a collaborative manner rather than in isolation.
3. Consistent methodological paths, with the inherent logic of evidence-based methods being the same—data diagnosis, targeted improvement, optimization, transitioning from data insights to classroom understanding and teaching action improvement, facilitating teachers’ transformation from experience-based to research-based practices.

While the features differ, the underlying logic is interconnected. This cross-regional transferability gives Li Yushun confidence in promoting this methodology.

What gives him even more confidence is the “UGBS” regional educational collaborative innovation model, termed the “Navigating and Integrating Model.”

In this model, the Beijing Normal University team is responsible for theoretical innovation, indicator system construction, and evidence-based method development; Seewo translates theory into actionable intelligent systems, achieving large-scale cloud deployment and normalized services. Both parties do not simply provide products but deeply integrate technology, theory, and teaching practices by embedding themselves in regions and schools. “Our goal is clear—using this opportunity to accelerate the transition of basic education classrooms from experience paradigms to data-driven, evidence-based new stages, making every regular class a site for diagnosable, traceable, and iterative teacher professional growth.”

At the 87th China Education Equipment Exhibition, Seewo launched its fifth-generation AI recording solution, completing a comprehensive upgrade of four major recording products: premium course production, regular teaching recording, lightweight deployment in ordinary classrooms, and flexible mobile shooting, all finding smarter and more professional solutions.

Seewo launched its fifth-generation AI recording solution.

“Education needs to be grounded and cannot be superficial or noisy,” Li Yushun stated, emphasizing that this pragmatic cooperation allows the project to take root nationwide.

Facing the 14th Five-Year Plan, the strategic opportunity for AI to empower education has already begun.

Li Yushun told reporters that building on existing achievements, Beijing Normal University and Seewo will continue to promote the universal application of results with a sense of responsibility and commitment to lead the field, accelerating the large-scale, normalized, and systematic transformation of Chinese classrooms. Both parties will also pay more attention to reconstructing the support system for professional development of teachers in county areas, delivering digital professional development services to teachers in their practical positions.

Undoubtedly, the significance of AI empowering evidence-based classroom development lies not only in making every class “evidence-based” but also in promoting educational practices from experiential pedagogy to scientific pedagogy. The future of AI + education is promising.