laitimes

Political, academic and enterprise guests discussed "big model production and industrial landing": big models must be applied to practical scenarios to empower social development

author:Cover News

Cover news reporter Wu Yujia

On July 8, the 2023 World Artificial Intelligence Conference "Large Model Production and Industry Landing Cooperation Forum" was held in Shanghai, guests from government, academia and enterprises conducted heated discussions on technology trends, model production, industrial applications and computing power needs, jointly promoted the exchange and cooperation of large model technology and sustainable development, helped the construction of a scientific and technological power, and contributed to the development of artificial intelligence in the world.

Gong Hongbing, member of the Party Working Committee of the Administrative Committee of Lingang New Area, said that artificial intelligence is an important driving force for the intelligent technology revolution and industrial transformation, and the development and application of large models will become the focus and core of accurate artificial intelligence technology change.

Political, academic and enterprise guests discussed "big model production and industrial landing": big models must be applied to practical scenarios to empower social development

So, how to play the role of the big model? He Jifeng, an academician of the Chinese Academy of Sciences, believes that from the technical route, we hope to have a more powerful computing power platform, which can allow us to do data analysis, collect more data, and produce larger models. We hope to establish an open ecological model, which has two meanings: on the one hand, we should put more platforms in the open source environment, so that everyone can use and experience, and improve the quality of this platform. On the other hand, we hope to build a good industry ecology, and large enterprises can drive small and medium-sized enterprises to make full use of artificial intelligence and empower their technological development. In addition to exchanging technical opinions at this forum, we will meet many new friends, and we hope that new and old friends will work together to do a good job in the work of artificial intelligence empowering the real economy.

Shi Yong, Counsellor of the State Council, Academician of the Academy of Sciences for Developing Countries, Academician of the International Eurasian Academy of Sciences, Director of the Research Center for Virtual Economy and Data Science and Technology of the Chinese Academy of Sciences, and Director of the Key Laboratory of Big Data Mining and Knowledge Management of the Chinese Academy of Sciences, analyzed the current situation and pain points of the big model industry, as well as the development opportunities and challenges that may be ushered in in the future. In Shi Yong's view, the mainland's artificial intelligence technology wants domestic technology rather than localization, so the development of mainland artificial intelligence has a long way to go.

Political, academic and enterprise guests discussed "big model production and industrial landing": big models must be applied to practical scenarios to empower social development

How to better move towards the AGI era, Shinma Intelligent has given its own answer. At this forum, Flash Intelligence introduced the ATOM AI productivity platform and announced the perception large model SupreMeta. Peng Yao, founder, chairman and CEO of Flash Horse Intelligence, said, "From the perspective of future needs, artificial intelligence is a kind of productivity. The essence of science and technology should return to solving problems, and we look forward to building an AI productivity platform to truly serve the development of 'AI+' and achieve a leap in the efficiency of urban space exploration." "Based on the ATOM AI productivity platform, Shinma Intelligent itself has also done a large model exploration." The significance of the existence of the big model must be applied to actual scenarios to solve problems for people and empower social development. For the large model itself, Shinma Intelligent combines its own advantages and makes a perception large model based on visual expansion. Peng Yao said. The perception large model SupreMeta of Shingema Intelligent unifies detection, classification, and segmentation for multiple tasks; Visual, radar, and laser multimodal fusion to achieve an overall effect greater than a single modal or single-task analysis. Compared with Meta's open-source visual large model SAM, its measured effect in the transportation field is more than 15% better.

Zhang Yingfei, a distinguished researcher at the Virtual Economy and Data Science Research Center of the Chinese Academy of Sciences, said: "Zhongke Shenma Intelligent Co., Ltd. has three advantages: first, strong alliance and resource integration; Second, technology integration and expansion of application; Third, focus on efforts and break through difficulties. We look forward to working with Flash Horse to promote the development and innovation of independent and controllable domestic technology. ”

Wu Fei, Qiushi Distinguished Professor of Zhejiang University, Executive Deputy Dean of Shanghai Advanced Research Institute of Zhejiang University, and Director of the Institute of Artificial Intelligence of Zhejiang University, shared the modes and methods of ChatGPT from general to vertical fields. In Wu Fei's view, "the art industry has a specialty, can you train a model for vertical field tasks, and now there are many models with hundreds of billions of parameters and trillions of parameters to identify, compress, quantify, and turn into smaller models, and use the data in this field to make it align and make a better model traction." ”

Zhang Xiaodong, chairman of Wuzhen Think Tank, shared the theme of "The First Principles of Large Models". Why is artificial intelligence hot again in 2023, ChatGPT is born is that there is a new breakthrough in the field of artificial intelligence - the breakthrough of language, "We agree that language is the most powerful thing for people, language is what distinguishes us from animals, there are many animals with better hearing than people, there are many animals with better vision than people, and language equals thinking, so this is the core of this revolution." ”

Shen Guohong, chairman of Shanghai Urban Construction Digital Industry Group, analyzed in detail the "AI-empowered Yangpu Bridge Digital Twin", which gave everyone a clearer understanding of the architecture of the digital twin of road transportation infrastructure. "As the hub of Pudong and Puxi, about 120,000 vehicles walk back and forth on the Yangpu Bridge every day, and what we have to do is to provide some help through the interpretation of the structure of the bridge and the ability of AI."

Junbo Zhang, CEO of Unity China, shared the theme of "Trends and Prospects, Generative AI Technology Helps Unity Creation". In Zhang Junbo's view, through AI learning tools, creators can exponentially improve their creative efficiency, so that they can have more energy to invest in the creative work itself, and enterprises can also use AI to create higher quality products at a lower cost and have the opportunity to develop new businesses.

Chen Weiliang, Founder, Chairman and CEO of Muxi, shared the challenges and opportunities of the new stage of AI. Chen Weiliang said that the outbreak of large models has given many industry development opportunities, the industrial chain of this industry must be very long, different enterprises play different roles, some do technology large models, some do vertical domain models. Large models have also led to a lot of applications, which ultimately improve production efficiency and improve the quality of human life.

Wang Xin, CTO of Shanghai Jiafeng Chelu Digital Technology Co., Ltd., shared the theme of "Coordinated Development of Traffic Model and Intelligent Road Network". Wang Xin said, "Vehicle-road coordination itself is a miniaturized ecology, we still have to focus on industrialization, more consideration to strengthen data-driven decision-making, because we know that now we have too many elements on the road to manage, from the perspective of manual and traditional data collation, it may not manage all aspects, we must put some traditional management processes to data-driven intelligent management, so as to improve the quality of our public services." ”

[If you have news leads, welcome to report to us, once adopted, there is a fee to thank you.] Report WeChat concern: ihxdsb, report QQ: 3386405712]