At the main forum on industrial development of the 2024 World Artificial Intelligence Conference, Baidu founder Robin Li delivered a thought-provoking speech. He clearly pointed out that in the era of artificial intelligence, developing "super capable" applications is more important than pursuing "super applications" with a daily active user number (DAU) of 1 billion. Robin Li emphasized that we should go beyond the thinking mode of the mobile era and avoid falling into the "super application trap" and realize that the definition of success should not be limited to the number of users. This view points out a new direction for the development of AI applications.
Robin Li proposed that the agent is the simplest form of developing AI applications and is also the direction of development of AI applications that he is most optimistic about. He pointed out that open source models have their value in the fields of academic research and teaching, but in commercial competition, closed source models are more applicable due to their efficiency and cost advantages. He explained that open source models require a larger parameter scale to compete with closed source models, which leads to higher inference costs and slower reaction speeds. In addition, the use of open source models may result in the inability to share computing power and benefit from the continuous upgrade of the underlying model. This analysis provides an important reference for enterprises in the selection of AI models.
Li Yanhong also shared Baidu's practical experience, mentioning that after the novel creation shifted from open source model to closed source model, the availability and quality rate of novel generation have been significantly improved. He believes that the value of a big model lies in its performance in specific applications, not just the model itself. This view emphasizes the importance of AI technology in practical applications, rather than simply pursuing model scale.
Regarding the importance of AI applications, Li Yanhong gave an example to illustrate that AI applications have processed orders through large models in the express delivery industry, which significantly improved efficiency. He also mentioned that Baidu's Wenxin Kuai Code is gradually penetrating in the field of code generation, and about 30% of the code within Baidu has been generated by AI. These examples demonstrate the widespread application and great potential of AI technology in various industries.
Robin Li predicts that with the enhancement of basic model capabilities, it will become easier to develop agents, and millions of agents will appear in the future to form a huge ecosystem. He took Baidu's college entrance examination agent as an example to illustrate the huge potential of agents in practical applications. This prediction portrays a broad prospect for the future development of AI technology.
Finally, Li Yanhong discussed whether AI will replace human work. He believes that AI is currently more used as an auxiliary tool than a replacement. The development of AI technology has created some new job opportunities, such as data annotators and prompt word engineers. He emphasized that AI is always just a tool, and its purpose is to meet human needs, enhance human abilities, and make life better. This view provides a positive and positive interpretation of the social impact of AI technology.