On June 30, 2023, the AICA Chief AI Architect Training Program jointly established by the National Engineering Research Center for Deep Learning Technology and Application and Baidu ushered in the opening ceremony of the seventh term. 98 corporate CTOs, scientists, and technical executives stand out from more than 500 applicants. They will carry out a half-year learning and practice of artificial intelligence cutting-edge technologies and applications, and become leading talents in the implementation of large-scale model industries in various industries.
Opening Ceremony of the 7th Phase of AICA Chief AI Architect Training Program
At the opening ceremony, Wu Tian, vice president of Baidu Group and deputy director of the National Engineering Research Center for Deep Learning Technology and Application, said in his speech that the strong versatility of artificial intelligence technology enables it to empower thousands of industries, and the key to industrial intelligence is In the general trend of the integration of artificial intelligence and the real economy, the original professional and technical talents in various industries can become new compound talents who use AI technology to solve practical business problems. AICA’s chief AI architect training plan is to train and deliver such talents for the industry, and at the same time provide them with handy tools, including the Flying Paddle deep learning platform and Wenxin large-scale model series, especially the large language model Wenxin can understand a word , Can write, have logic, have emotion, and will also give full play to the advantages of logical thinking, reasoning and decision-making in the process of deep integration with the industry, bringing new value imagination space.
According to reports, Wenxin has provided 37 large models, covering natural language processing large models, visual large models, cross-modal large models, biological computing large models, etc. Based on the joint optimization of flying paddles and Wenxin large model, Wenxin large model has been iterated to version 3.5. Compared with version 3.0 in March, the training speed has increased by 2 times, the inference speed has increased by 17 times, and the cumulative effect of the model has increased by more than 50 %
Wu Tian, vice president of Baidu Group and deputy director of the National Engineering Research Center for Deep Learning Technology and Application
Since the first phase of the AICA project was launched in 2019, 322 AI architects have been sent to the industry. The seventh session can be described as the first training course for AI architects in the era of large-scale models. From the background and topics of the students, we can see the deep influence of large-scale models. On the one hand, the large model has become a new engine of industrial intelligence, leveraging the deep integration of intelligent technology and the real economy. The background experience and research topics of the students in the 7th class are cross-integrated and expanded towards a more applied and engineering direction. On the other hand, large language models, multimodal large models, and industry large models have emerged in large numbers. How to transform the technological dividends of large models into market competitiveness has become an urgent issue for enterprises.
71% of the students in the 7th phase of AICA came from central state-owned enterprises such as State Grid, Three Gorges Corporation, ICBC, and Agricultural Bank of China, as well as listed companies such as Shell, Haier, Midea, and China Merchants Bank. For the first time, 21 core first-level application service providers in the industry were also admitted. In addition to energy, intelligent manufacturing, and finance, the trainees have added a large number of leading companies in the fields of automobile transportation and corporate services, such as Volkswagen, BMW, Geely, BYD, China Southern Airlines, UFIDA, and Layton Automobile. Among the trainees, senior technical managers such as corporate CTOs and scientists accounted for 93.2%, a new high for AICA.
It is understood that in terms of the academic system, Baidu has set up a three-teacher teaching mode of "expert lectures, full-time teaching assistants, and group discussions" for AICA students. In the seventh phase, a "specially hired expert group" was added for the first time. The first batch of 12 specially hired experts include industry experts. Leaders, well-known scholars, university experts, and experts with outstanding achievements in specific fields not only have a high level of knowledge in the field of artificial intelligence, but also enjoy a high reputation and influence in the industry. They will provide students with unique industry insight, practical experience and interpretation of the latest technology trends. This will help enhance the professional quality and competitiveness of the trainees, and lay a solid foundation for their career development.
Under the wave of industrial intelligence, the insights of industry experts are becoming more and more important. As a representative of the specially-appointed expert group, Fang Fang, vice president of North China Electric Power University, shared at the opening ceremony the energy industry's demand for artificial intelligence technology, the cooperation between North China Electric Power University and Baidu in AI talent training, and some future-oriented projects based on large models. Exploratory thinking resonated with students with backgrounds in the electric power industry.
At the opening ceremony, 12 former AICA outstanding students from Beijing Telecom, China Property Reinsurance, Sunyard, China Electronics, iSoftStone, NI China Instrument, etc. became "flight assistants" and will provide assistance to the students of AICA 7th class , to jointly promote the new development of AI.
The large model accelerates industrial intelligence, and puts forward higher and newer requirements for compound AI talents. Ma Yanjun, general manager of Baidu AI technology ecology, pointed out in the theme sharing that AI architects must have the capabilities of business, algorithm, and architecture. Now they will focus on the three core elements of application refactoring, large-scale model technology, and large-scale model suites. One is to combine the large model with one's own business, better do application refactoring, and give full play to the capabilities of the large model; the second is to master the key technologies behind the large model, such as pre-training and fine-tuning, so that you know what it is and why it is; the third is to master More useful, smoother, and more efficient tools to complete the data processing, fine-tuning, compression, deployment, and sceneization of large models, so that the large models can be truly used.
At the end of the activity, Yu Dianhai, Chief Architect of Baidu Flying Paddle, brought the students the first lesson of AICA "Recognition of Deep Learning". He explained the essence of deep learning technology in depth from three perspectives: the historical evolution of AI technology, the new paradigm of large models, and the implementation of business applications, and pointed out that it is necessary to use AI native thinking to drive the transformation of technology products and reconstruct the technical architecture and product system.
Global AI reengineering driven by large models is sweeping across all walks of life on a wider and deeper level. In this unprecedented productivity change, Baidu AICA Chief AI Architect Training Program is committed to helping individuals become talents, helping enterprises to cultivate and gather talents, activating Chinese enterprises with talents, grasping the dividends of the era of large-scale models, and continuously creating immeasurable growth volume value.
Follow 【Flying PaddlePaddlePaddle】public account
Get more technical content~