"The STAR Market Daily" reported on July 29 (by Intern Reporter Wu Yutong and Reporter Li Jiayi) that the 2025 World Artificial Intelligence Conference (WAIC) was grandly held in Shanghai from July 26 to July 28. This year’s conference, themed “Intelligent Era, Shared Future,” brought together over 800 global companies and showcased more than 3,000 cutting-edge exhibits covering core areas such as large models, AI terminals, and intelligent robots. The conference aimed to foster consensus through cross-disciplinary dialogues and promote the global artificial intelligence industry’s healthy and orderly development toward beneficial, safe, and equitable directions.

During the conference, Laura (位晨), Executive President of TAL Education Group, joined the live broadcast hosted by Cailian Press to discuss the profound impact of artificial intelligence and large models on the education industry. She noted that large model technology has resolved the “impossible triangle” of personalization, scalability, and high quality in education, paving the way for the realization of truly individualized teaching. Every student can now have their own AI learning companion, and every teacher will be equipped with an AI tutor assistant. AI empowers educational scenarios from multiple dimensions, while its core remains focused on serving the essence of education—"nurturing human potential."

AI Breaks the 'Impossible Triangle in Education' and Builds a New Paradigm for Personalized Learning

Laura pointed out that the education industry has long faced the contradiction of achieving scalability, personalization, and high quality. For instance, one-on-one teaching delivers personalization and high quality but is hard to scale, while large-class lessons can accommodate more students but often fail to address individual differences. The application of artificial intelligence has offered a new solution to this “impossible triangle” in education.

Xueersi’s intelligent learning companion “AI Thinkie” (小思) provides personalized instructional explanations tailored to each child’s unique circumstances, truly implementing individualized teaching. Laura highlighted that such AI companions are gradually becoming regular aides in children’s learning. Meanwhile, teachers are now supported by AI assistants that enhance efficiency in lesson preparation, grading, Q&A, and test creation processes.

MathGPT: Constructing a Vertical AI Base for Education

Discussing TAL’s efforts in the large model space, Laura introduced TAL’s self-developed vertical model for education, "MathGPT" (九章大模型), which is among the first educational models to gain official national certification. Compared to general-purpose large models functioning as "all-rounders," vertical education models are more like "specialists," with deep insights into knowledge systems and student learning paths within teaching scenarios, making them better suited for the complex and layered field of education.

The MathGPT model has already seen widespread application across TAL’s learning tablets, classroom products, and intelligent education materials, achieving a closed-loop application of AI capabilities in teaching. By incorporating multimodal data—including question banks, test papers, videos, and books—in training, the model aligns closely with real teaching and learning needs.

From 'Content + AI' to 'Human-AI Collaborative Teaching'

On the core competitiveness of AI in education, Laura emphasized TAL’s advantage in "Excellent Content + Advanced AI." Over the past 22 years, the company has amassed comprehensive learning resources across all subjects and grades, spanning in-school courses, advanced extension learning, and quality education dimensions—all of which are integrated into its AI system.

In current classroom applications, TAL has established a "four-screen interactive intelligent teaching system," comprising a teacher's screen, student devices, a teacher’s tablet, and a parent interface, facilitating real-time synchronization of teacher-student interaction and parent-teacher communications. Laura believes leveraging AI to enhance instructional efficiency and interactivity is key to modernizing education.

Addressing concerns about whether AI might replace teaching roles, Laura firmly stated, “Teacher positions cannot be replaced.” She noted that education is not just about transmitting knowledge; it’s also about emotional connections and character development. AI can handle the repetitive tasks in teaching, easing the burden on educators, but the essence of education—like empathy, encouragement, and role modeling—is uniquely human and irreplaceable.

Laura revealed that TAL’s MathGPT development team operates on a high-frequency iteration schedule while maintaining ongoing international dialogues. A recent visit by delegations from 12 countries’ education ministries left a strong impression of their AI educational products and rapid innovation cycles. She expressed that TAL will continue to increase investments in education technology, advancing AI as a critical tool in global foundational education.

Source: Cailian Press https://www.cls.cn/detail/2099583