On May 31, the [Creating Intelligence·Exploring the Unknown] Hangzhou General Artificial Intelligence Forum - AI basic software cutting-edge technology analysis was hosted by the China Academy of Information and Communications Technology and the China Artificial Intelligence Industry Development Alliance, and co-sponsored by Beijing Jiuzhang Yunji Technology Co., Ltd. The forum was successfully held in Hangzhou. The conference focuses on the cutting-edge technology of AI basic software, explores the new growth brought by artificial intelligence basic software + large models to the artificial intelligence industry, discusses the challenges and opportunities faced by AI engineering in practical applications, and effectively promotes the integrated development of my country's AI infrastructure. .
At the same time as the conference, a signing ceremony was held between the China Academy of Information and Communications Technology’s Yun Da Institute and Jiuzhang Yunji DataCanvas Company to deepen the cooperation on "industrialization of high-quality AI infrastructure." The two parties will have extensive cooperation in standard formulation, evaluation and evaluation, technological innovation, and industrial research. On the basis of this, they should give full play to their respective resource advantages in theoretical research, technological innovation and application practice in AI infrastructure, open up the upstream and downstream ecological chains of AI infrastructure, and jointly build an open, powerful and flexible AI infrastructure ecosystem. Wei Kai, deputy director of the Cloud Institute of the China Academy of Information and Communications Technology, said that Jiuzhang Yunji DataCanvas Company and the Cloud Institute of the China Academy of Information and Communications Technology have once again joined hands to focus on the construction of a large model industry system and accelerate the application of large models in vertical industries such as finance, communications, and transportation. application empowerment.
China Academy of Information and Communications Technology's YunDa Institute and Jiuzhang Yunji DataCanvas Company deepened the cooperation signing ceremony on "industrialization of high-quality AI infrastructure"
A new era of large models has begun, and the core position of AI basic software is highlighted
Shang Mingdong, co-founder of Jiuzhang Yunji DataCanvas, said at the meeting that large models will present diversified development paths in the future. "Computing power, data and basic software are the three important factors that influence the development of large model diversification paths." In the future, as computing power performance gradually becomes homogenized and standardized, the differences in data and the personalization of enterprise needs will gradually increase. “AI basic software” will become a decisive factor in model training efficiency and computing power usage efficiency. As the backbone of the model ecosystem, AI basic software will become the most important efficiency support for the implementation of large model applications, and form a new paradigm for model training through the combination of large models + small models.
Shang Mingdong, co-founder of Jiuzhang Yunji DataCanvas Company
"The massive multi-modal data management at the bottom and the need for more accurate analysis and decision-making at the upper level will push the integration of digital intelligence into deep waters and bring new opportunities for building AI basic software." Li Haoran, senior analyst of IDC China's Artificial Intelligence and Big Data It is predicted that for the basic software of development service platform that customers are more concerned about, technology companies should build it from six aspects: full life cycle components, low code/no code, automatic machine learning, algorithm model library, visualization, deployment and operation, and Focus on integration with cloud services and big data components.
IDC China Senior Analyst of Artificial Intelligence and Big Data Li Haoran
Technology application is the ultimate goal, and thousands of industries are accelerating the large-scale application of AI
Regarding the key construction of the six major aspects of AI basic software proposed by IDC, Zhou Xiaoling, vice president of Jiuzhang Yunji DataCanvas, said that the company has long-term layout of these important technical capabilities and applies them in finance, communications, In industries such as transportation, manufacturing, and energy. He said that the development path of AI technology from decentralized models to integrated intelligence and then to general artificial intelligence has greatly promoted the wave of digital intelligence in governments and enterprises; each industry has its own development characteristics and transformation stage differences. In cloudification There is considerable demand for iterative upgrades of AI platforms and AI application capabilities in technical fields such as automation, multi-modality, and distribution. There is still huge room for development of AI applications from operations to operations.
Zhou Xiaoling, Vice President of Jiuzhang Yunji DataCanvas Company
The banking industry is a leading industry in digital transformation and digital intelligence upgrades, and it is also the first test of the large-scale availability of cutting-edge technologies. Zhao Chenguang, a technical expert from the Retail Finance Department of Hangzhou Bank, delivered a keynote speech on "Sellers and Buyers of AI Banking" at the meeting. From the perspective of actual users of AI technology, he shared insights on the application of cutting-edge artificial intelligence technology in multiple front-line business scenarios of banks. He pointed out that the AI pyramid built with the model layer, data layer and application layer provides a solid foundation for AI banks, which will build smarter, more efficient and more accurate operations and service capabilities for banks in the future.
Technical expert of Hangzhou Bank Retail Finance Department Zhao Chenguang
Independent innovation is the source of power, open source and openness are the driving force
As the data and demand environment in the digital upgrade process become more and more complex, governments and enterprises have put forward higher requirements for the scientificity and efficiency of decision-making based on data. Yang Jian, chief architect of Jiuzhang Yunji DataCanvas Company, said that in this context, artificial intelligence technology represented by decision-making intelligence has become an indispensable technical support in digital intelligence upgrades. The Jiuzhang Yunji D-lab open source team has been continuously exploring and accumulating in the field of decision-making intelligence based on structured data for many years, and has released more than ten open source projects related to automatic machine learning, deep learning, causal learning and other related technologies. Seeking to empower governments and enterprises to better realize data-driven automated and intelligent decision-making. Under the technological upsurge of AIGC, the Jiuzhang Yunji D-lab open source team is carrying out cross-type research to accelerate the integration and innovation of cutting-edge AI technologies.
Yang Jian, Chief Architect of Jiuzhang Yunji DataCanvas Company
Industry drives technology, and technology empowers industry. At present, the scale of industrial application of artificial intelligence technology is getting larger and larger, and its operations are becoming more and more complex. Hu Hui, an engineer at the Cloud Institute of China Academy of Information and Communications Technology, said at the meeting that MLOps will solve the "difficulty in production, sustainability, and difficulty" faced by AI engineering management. problems such as reproducibility and difficulty in collaboration. MLOps not only allows users to significantly reduce costs and increase efficiency in AI applications, but also has emerging popular professions such as MLOps engineers. The future development trend is evident. In 2023, the China Academy of Information and Communications Technology will release the first domestic MLOps practice guide. Together with the two previously released MLOps standards and the upcoming series of standards such as the large model operation LMOps standard and the model operation standard, the China Academy of Information and Communications Technology will implement MLOps technology. Make think tank contributions to industry standardization and application scale.
Engineer Hu Hui from the Cloud Institute of China Academy of Information and Communications Technology
On the same day, the China Academy of Information and Communications Technology released a series of standards for AI development platforms, among which Jiuzhang Yunji DataCanvas Company was deeply involved in the "General Capability Requirements for Artificial Intelligence Development Platforms Part 2: Security Requirements" and "Technical Requirements and Test Methods for Automatic Machine Learning Platforms" The formulation of four standards, "Part 1: Platform Functions", "Technical Requirements and Testing Methods for Machine Learning Platforms for Structured Data Part 1: Functional Requirements", "Technical Requirements and Testing Methods for Deep Learning Platforms Part 1: Platform Functions", and Obtained the license from the core participating units.
The current wave set off by ChatGPT and AIGC has had a huge impact and actual demand in the industry. The large-scale emergence of generative AI is expected to make 2023 a critical year for artificial intelligence to truly empower thousands of industries. IDC predicts that by 2026, large-scale base models will become a standard industry utility offered by large vendors. AI basic software will write the answer to self-reliance and technological innovation driven by the times and industry.