Source: Wall Street News
In the midst of increasingly fierce competition, Baidu is striving to maintain its leading position in the Chinese AI market. Wenxin said that the number of users exceeded 200 million, doubling from the end of last year.
On Tuesday, April 16, the Create 2024 Baidu AI Developer Conference with the theme of “creating the future” was held at the Shenzhen International Convention and Exhibition Center. Baidu CEO Li Yanhong pointed out at the conference that Wenxin currently has 200 million users, double the scale from December last year, and the average number of daily API calls has also exceeded 200 million.
At the same time, Li Yanhong pointed out that Wenxin had more than 85,000 corporate customers. Last month, the media quoted people familiar with the matter as reporting that Apple and Baidu had preliminary negotiations on the use of Wenxin in the Chinese market.
In an earnings conference call in February of this year, Baidu said that the technology had begun to generate revenue for the company. In the fourth quarter of last year, the company used AI technology to improve advertising services, help other companies establish turnover models, and earned hundreds of millions of yuan.
Baidu expects AI's revenue contribution to be more significant in 2024, while its core business will remain steady.
Currently, rhetoric is also facing fierce competition. Over the past year, big models and AI chatbots have sprung up endlessly, including Alibaba's Tongyi Thousand Questions, Tencent's “Hybrid”, and Kimi, an intelligent assistant product under Dark Side of the Moon supported by Alibaba.
According to Barclays forecasts, China's artificial intelligence market is expected to grow dramatically, exceeding US$26 billion by 2026, a significant increase compared to less than US$15 billion last year.
At this AI developer conference, Baidu also brought three development tools: AgentBuilder, AppBuilder, and ModelBuilder.
According to Li Yanhong, AppBuilder has pre-packaged and pre-configured the various components and frameworks needed to develop AI native applications, drastically lowering the development threshold. In as fast as three steps, developers can use natural language to develop an AI native application and easily publish it in various business environments.
AgentBuilder is based on a powerful basic model. Agents can be generated in batches and applied in various scenarios. Li Yanhong said at the conference:
Up to now, more than 30,000 smart devices have been created, more than 50,000 developers, and tens of thousands of enterprises have settled in.
Li Yanhong also shared ModelBuilder, a tool for customizing models of various sizes. ModelBuilder has “the ability to produce models efficiently and at low prices”. He interacted with Xiaodu in real time on site, showing that Xiaodu uses multiple model combinations to perform different tasks.
For example, the small model ERNIE Tiny is used to perform model routing work, while Wenxin 4.0, which has the best performance, is used to perform complex requirements such as scheduling. According to reports, compared to the flagship version, which uses all of Wenxin's big models, Xiaodu can increase the response speed by 2 times and reduce the cost by 99%.
In addition, Baidu Executive Vice President Shen Dio officially released the next-generation intelligent computing operating system, Wan Yuan. By abstracting and packaging the intelligent computing platform in the AI native era, the system blocks the complexity of cloud-native systems and heterogeneous computing power for users, and enhances the efficiency and experience of AI native application development.
Shen Diao said:
As big model technology continues to evolve, programming through natural language is becoming a reality. Programming will no longer be process-oriented or object-oriented, but demand-oriented; the process of programming will become a process for developers to express their wishes and bring revolutionary changes to the operating system.
In the core of the operating system, the underlying hardware changed from mainly CPU computing power to mainly GPU computing power, and added world knowledge compressed by large models. The object of operating system management has fundamentally changed, evolving from managing processes and managing microservices to management intelligence.
Editor/jayden