Recently, the award ceremony for the 2024 "Data Element×" Beijing Division Competition concluded successfully at the Zhongguancun International Innovation Center. The project "Key Technologies and Applications of Intelligent Evaluation and Multi-Level Precise Guidance Assisted by Large Language Models," jointly developed by TAL Education Group and Jinan University, won the championship in the Smart Education track.

The "Data Element×" competition is the first national competition focusing on the development and application of data elements in China.  The aim is to promote the development of relevant technologies and industries by selecting a batch of data element development and utilization solutions with significant application results, strong innovation, and leading effects. Since its inception in June this year, the Beijing Division received 729 project entries from across the country, including from data commerce companies, renowned universities, research institutes, and state-owned enterprises, covering 16 tracks such as technological innovation, financial services, industrial manufacturing, and smart education. After several rounds of selection, the Beijing Division awarded a total of 32 first prizes, 45 second prizes, and 73 third prizes.

Leveraging its leading educational philosophy and advanced technological capabilities, TAL, along with Jinan University, developed the "Key Technologies and Applications of Intelligent Evaluation and Multi-Level Precise Guidance Assisted by Large Language Models" (hereinafter referred to as Intelligent Guidance Technology). This project won the first prize in the Smart Education special competition, supported by the National Next Generation Artificial Intelligence Open Innovation Platform for Smart Education and the Guangdong-Hong Kong-Macao Smart Education Joint Laboratory.

The Intelligent Guidance Technology integrates advanced technological capabilities such as large language models, artificial intelligence, big data, and intelligent knowledge graphs, deeply exploring and utilizing the potential of large language models and related AI technologies in smart education.

Historically, the education sector has faced the "impossible triangle," which means it has been difficult to achieve personalization, high quality, and large scale simultaneously. The emergence of large language models offers a solution to this problem. These models can deeply understand and generate language, providing a more natural interactive experience. On this basis, the Intelligent Guidance Technology builds more precise and enriches educational knowledge graphs, realizes multi-level、fine-grained knowledge structuring and multi-level deep cognitive diagnosis, ultimately providing students with more precise and personalized learning content.

Currently, TAL has launched the intelligent question-answering math tool "MathGPT Anytime Ask" based on the Intelligent Guidance Technology. This tool helps students solve math problems through heuristic guidance, aiming to cultivate their problem-solving thinking and improve their math learning ability.

When students use "MathGPT Anytime Ask" to address the math problems which they encounter in their studies, it does not provide answers directly but first analyzes the knowledge points and forms being tested by the problem. Then it guides students to think step by step about how to solve the problem, and heip them gain inspiration through exploration. This method not only helps students better understand the problems but also develops their ability to solve problems independently.

As the builder of the National Next Generation Artificial Intelligence Open Innovation Platform, TAL has always attached great importance to investment in technological innovation. As early as 2017, it established the "AI lab (AI Laboratory)," which is the first post-doctoral base in the education and training industry and launched the country's first large-scale math model with a hundred billion parameters—the MathGPT.

"Education as the foundation, data empowerment," TAL will continue to deeply explore the potential of large language models and related AI technologies in smart education, promote educational transformation and innovation, meet the educational needs of different learning levels and states, and contribute to the development of the education sector.