Recently, at the Beijing AI Innovation Source Promotion Conference, the Beijing Municipal Science and Technology Commission and the Zhongguancun Administrative Committee released the first batch of Beijing AI application scenario joint R&D platforms. These platforms cover key industries including government service, education, smart city, culture and sports, industry, and finance, exploring new paths for the deep integration of large model technology and application scenarios.

Among them, the Foundational Education Beijing AI Scenario Application Platform, guided by the Beijing Municipal Education Commission, is jointly built by Beijing Normal University and TAL Education Group. At the event, Huang Hua, Dean of the School of Artificial Intelligence at Beijing Normal University, and Tian Mi, CTO of TAL Education Group, signed on behalf of both parties to officially launch the platform.

The Foundational Education Beijing AI Scenario Application Platform (hereinafter referred to as the Application Platform) aims to build an educational large model in the field of foundational education. Based on this, it will eventually form a related application ecosystem and generate a series of solutions, overcoming the "last mile" in the implementation of AI applications. In recent years, the research on foundational education large models has gained widespread attention and achieved certain progress. However, there still exist issues such as a lack of ecosystems and comprehensive application solutions. The complexity of education and academic knowledge makes it difficult for traditional AI products, which lack complex processing and deep understanding capabilities, to provide high-quality services in the field of foundational education. Therefore, it is urgently needed to develop key technologies for large models characterized by foundational education under the guidance of educational and teaching theories.

According to reports, the Application Platform will focus on breakthroughs in technologies such as evaluation methods for multimodal foundational education large models supporting full-scenario integration, construction and cleaning methods of data in the field of foundational education, and efficient training strategies for foundational education base large models. Ultimately, it aims to develop referenceable technical specifications, an open evaluation platform, open-source base large models, and foundational education application solutions based on large models. The Application Platform will be demonstrated in primary and secondary schools to support the testing ground for large models in the Beijing education field, provide evaluation services, accelerate the development progress of enterprise application solutions, reduce R&D costs, assist in the ecological construction and healthy development of the foundational education large model industry, and enhance the intelligence level of the foundational education industry.