This article originated from "jazzyear", ID：jazzyear

On April 26, "Technology returns to the city, talents create excellence" - the Haishu District Entrepreneurship Conference of 2023 was grandly held at New Century Grand Hotel in Zhongdu by the Yongshui River. The event was sponsored by the CPC Committee and People's Government of Ningbo Haishu District, and jointly organized by the Haishu District Talent Office, Haishu District Science and Technology Bureau (Association for Science and Technology), and Alibaba Cloud.

At the event site, famous experts gathered and people actively paticipated with no empty seats. He Jifeng, academician of the Chinese Academy of Sciences and dean of Shanghai Huake Zhigu AI Research Institute, delivered a keynote speech on AIGC. Experts from industry and academia, Huang Xiaoqing, Liang Xiaoyao, Xu Tao, Liang Yu and other experts shared their understanding of generative AI, AIGC, meta research and views on emerging concepts such as the universe. Lanzhuo Digital Technology Co., Ltd., a representative enterprise of innovation and entrepreneurship in Haishu District, explained how the local industry in Ningbo uses technology to break through the future. Zhang Yijia, the founder and CEO of "jazzyear", and six guests discussed large models and the future of business in the "AI Revolution's New Era" at a round table.

As one of the main organizers, Alibaba Cloud specially created a virtual digital human named Xiaoyun for the event. With rich movements and realistic demeanor, Xiaoyun not only presided over the entire conference, but also released the "Shangyun Magic Cube" manual at the end of the conference to help small and medium-sized enterprises lower the threshold of digital transformation.

Lectures by academicians and relays by big names to discuss the future of generative AI

Generative AI is the most concerned topic in the technology industry this year, and the products of domestic and foreign technology giants are dizzying. The first part of the conference focuses on the progress of technological industrialization in the new era of AI.

The voice of the academic community on generative AI is still rarely heard. At this conference, He Jifeng, academician of the Chinese Academy of Sciences and president of Shanghai Huake Zhigu AI Research Institute, first gave a speech on "AI Empowering Business Innovation". Academician He made his remark that data, algorithms, and computing power need to develop simultaneously with the rapid development of AI. Based on this, giants and small and medium-sized enterprises in various fields such as data intelligence companies, digital service companies, and cloud services need to work together.

Xu Tao, general manager of Dingding in Zhejiang Province, has a view on generative AI that "it seems to be a blockbuster, but it has been planned for a long time." On April 18, at the 2023 Spring Ding Summit, Dingding posted a slash "/", and demonstrated on-site access to the Qianwen (i.e. thousands of questions) large model, and input "/" to invoke more than 10 AI capabilities on Dingding. In addition to IaaS and PaaS based on Alibaba Cloud, DingTalk will also shift from SaaS to MaaS (Model as a Service).

In addition to office scenarios, industry is considered one of the most suitable scenarios for generative AI. Dataa Robotics has just set up a branch in Ningbo. According to Huang Xiaoqing, the founder, chairman and CEO of Dataa Robotics, who is also the former dean of China Mobile Research Institute, "cloud-edge-terminal integration is the future of intelligent robots." Robots are the key, and Dataa Robotics is using multi-modal large models to build a general AI for robots.

Investors are also an indispensable driving force to push technology forward, and they are even one of the first groups to sense the changes in the industry. Liang Yu, partner of founding partner CCV, brought his research insights into AIGC at the conference. He believes that "data is the core competitiveness of the new era of AI", and the progress of new language models brought by these data will bring new opportunities.

With a good prospect, there are still some difficulties to be overcome in the development of large models and AI in China, and computing power chips are one of them. "Human development is the evolutionary history of computing power development." According to certain media reports, training ChatGPT requires 10,000 A100 chips, and training a large model usually requires millions to tens of millions of dollars. Under the influence of geopolitics, China must rely on self-developed chips to develop its own large-scale model, generative AI, and metaverse.

Professors from the Computer Science and Engineering Department of Shanghai Jiaotong University and the Advanced Computer Architecture Laboratory once created a GPU open source platform "Blue and White Porcelain". Liang Xiaoyao, director of the laboratory, shared in this event that "domestic GPU companies do not possess strong landing capabilities, mainly due to the limited development time of the software ecology." And this is exactly an open-source software ecological platform for entrepreneurs to reduce the cost of software ecological development, so that domestic chips can be implemented faster.