Beijing Launches World's Largest AI Model Training Platform

From June 1 to 3, Beijing hosted the Zhiyuan Conference, a leading international academic event centered on artificial intelligence (AI). Over 30,000 professionals from the AI sector registered for this year's conference, marking its third edition since its launch in 2019. The event offered both online and in-person attendance options.

The rise of large-scale models is crucial for advancing AI, providing a framework for transformative industry infrastructure. A major highlight of this year's conference was the introduction of "Wudao 2.0," China's first ultra-large-scale intelligent model. Tang Jie, academic vice president of the Beijing Zhiyuan Research Institute, noted that this model can replicate essential human functions, tackling key challenges in developing strong AI—such as answering questions, composing poetry, and generating images. Wudao 2.0 has the potential to significantly enhance traditional businesses and IT companies by boosting their computational capabilities and applications.

Zhang Hongjiang, chairman of the Beijing Zhiyuan Research Institute, underscored the importance of Wudao 2.0, calling it the largest AI model globally, with a preeminent number of parameters. This advancement showcases China's latest achievements in AI internationally. The conference also introduced "Hua Zhi Bing," China's first original virtual student developed from Wudao 2.0.

The Zhiyuan Conference hosted 29 thematic forums covering vital topics in global AI research and industry trends, including "Pre-trained Models," "Machine Learning," "Collective Intelligence," and "AI Entrepreneurship." These sessions ranged from foundational AI research to cross-disciplinary integration and industrial applications, offering a valuable platform for scholars, engineers, and technical experts to collaborate and drive innovation.