Recently, the 11th Internet Security Conference (ISC 2023) was grandly opened in Beijing. The conference is organized by the Organizing Committee of the ISC Internet Security Conference, the Internet Society of China, the China Cyberspace Security Association, the Big Data Operation and Maintenance (Network Security) Committee of the All-China Federation of Industry and Commerce, the Chinese Association for Artificial Intelligence, the China Software Industry Association, the China Enterprise Confederation, and 360 Internet. Hosted by the Security Center; co-organized by the China Association of Communications Enterprises and the China Association of Small and Medium Enterprises; organized by the Computer Security Professional Committee of the China Computer Federation and the Artificial Intelligence and Security Professional Committee of the China Artificial Intelligence Society.
With the theme of "Security as a Service, Opening a New Paradigm for Digital Security in the Era of Artificial Intelligence", this conference will create the world's first AI digital security summit, gather global security pioneers, expert think tanks, and industry leaders to participate deeply, and outline digital security and AI technology Integrating the new blueprint for development, Zilliz, as a leading company in the industry, was invited to participate in this conference, and discussed with all parties about the practical application and challenges of large-scale model implementation.
Guo Rentong, partner and product director of Zilliz, shared a theme titled "Memory of Large Models" in the lecture session of the on-site data experts. Guo Rentong started from the "AI illusion" of the large model. He said that the current "AI illusion" is one of the main pain points of large models. Building a knowledge base is an effective way to deal with this problem. The vector database can be used as an extension of the knowledge base to enhance the knowledge of large models. This also provides a new paradigm for building data applications in the LLMs era - CVP Stack. Among them, C is a large model represented by ChatGPT; V stands for Vector Database; P stands for Prompt engineering. Among them, C serves as a computing unit, providing logic analysis and natural language docking capabilities, V serves as a storage unit, providing stable, accurate, and high-capacity knowledge, and P provides specific business-oriented adaptation capabilities based on the former two.
In the whole CVP Stack, the vector database plays an important role, its