As big model technology gradually enters its mature stage, where is the breakthrough point for the AI industry?
On the morning of November 12, Robin Lee, the founder of Baidu, delivered a speech on the theme of "Applications Are Here" at the 2024 Baidu World Conference. He stated in his speech that intelligent agents and industrial applications are the two key directions for the development of AI applications. We believe that intelligent agents are the most mainstream form of AI applications and are about to reach its tipping point
Robin Lee said that today, all the top technology companies in the world pay close attention to agents, but few like Baidu regard agents as the most important strategic direction.
Robin Lee said that being an agent is much like being a website in the PC era or an We Media account in the mobile era. The difference is that agents are more intelligent, more like your sales, more like your customer service, and more like your assistant. Agents may become a new carrier of content information and services in the AI native era.
The low entry threshold but high "ceiling" of functionality is the main reason why intelligent agents are emerging in a large number of industries.
How to implement large-scale models in the industry is one of the important directions that Baidu continues to explore. Robin Lee said that in the past year and a half, in the fields of finance, energy, education, recruitment, public services, etc., the combination of big models and scenarios has achieved tangible results in both cost reduction and efficiency increase.
He said that at present, on the stage of Qianfan model of Baidu AI Cloud, he has helped users fine tune 33000 large models and developed 770000 enterprise applications.
In his speech, Robin Lee also released iRAG, a retrieval enhanced Wensheng Graph technology. This technology combines Baidu's billion level image resources with powerful basic modeling capabilities to generate various hyper realistic images. iRAG technology greatly improves the usability of AI generated images, with commercial value such as no illusion, hyper reality, no cost, and immediate availability.
Robin Lee also revealed that as of the beginning of November, the daily average adjustment volume of Baidu Wenxin Model had exceeded 1.5 billion, an increase of about 30 times compared with the 50 million times disclosed for the first time a year ago. Robin Lee said that "the growth rate exceeded expectations". He sighed that this steep growth curve represents the explosive application of large-scale models in China in the past two years.
Upstream News Reporter Yang Xinhua