Yangtze River reporter Wan Ling, correspondent Zhang Shanni, Li Yifan
City Observation Group: Last year's ChatGPT and this year's Sora sparked heated discussions as soon as they were released, and also set off a new round of artificial intelligence development trends. I want to ask Xiao Chu, what do you think are the connections and differences between you and them?
Digital person "Xiao Chu": The emergence of these two friends amazed many people, including me. ChatGPT is an AI chatbot, Sora is a video generation model, and when it comes to the same points, we are both AI models. We also share the same goal, which is to make people's lives easier. The difference is that I can not only chat but also generate videos, which combines the advantages of two friends, and can also support comprehensive Q&A tasks such as 3D understanding and signal analysis. This year, I will further upgrade and grow to 3.0, and I will tell you more about it when the time comes.
Wang Jinqiao, President of Wuhan Institute of Artificial Intelligence, introduced the Zidong Taichu model. Photo by Yangtze River reporter Peng Nian
This question and answer occurred at the special report meeting of "Wuhan New City, World Optics Valley" of the theme report meeting of "Accelerating the Transformation of 'Three Advantages' and Reshaping the 'Important' of Wuhan in the New Era" held on June 27. At that time, Wang Jinqiao, president of the Wuhan Institute of Artificial Intelligence (hereinafter referred to as Wuzhi Institute), was in charge of the live demonstration, and he was the only guest who brought a digital human to give a report.
Xue Yeshan, a member of the City Observation Group, asked the digital human "Xiao Chu". Photo by Yangtze River reporter Peng Nian
"Xiaochu" is a digital human bred by the large model of Zidong Taichu. Wang Jinqiao introduced that the Zidong Taichu model has continuously broken through technical barriers from the single research of graphics, texts and sounds in 2008 to the joint research of multi-modality, and has completed the iterative upgrade from version 1.0 to 2.0, adding video, signal, 3D point cloud and other modalities on the basis of voice, image and text.
Following the development of the "Zidong Taichu" multi-modal large model version 2.0 and the Zidong Taichu artificial intelligence open service platform, Wuzhiyuan also released the "Multimodal Basic Large Model Technology White Paper" and artificial intelligence application products such as "Jiangcheng Taisu", "Jiangcheng Luoshen" and "Variety Wukong". "In the future, everyone will be a content creator, and artificial intelligence will empower thousands of industries." Wang Jinqiao said.
This month, the intelligent weld guidance and identification detection system jointly developed by Wuzhi Institute and Huagong Technology was shortlisted for the first batch of typical cases of artificial intelligence-enabled manufacturing transformation and upgrading in Wuhan. Wang Jinqiao said that the intelligent weld guidance and identification detection system uses the "Zidong Taichu" multi-modal large model as the intelligent base, completes the collection of welding material data in a few seconds, and "stitches" an irregular stainless steel weld at a speed of 10 meters per minute.
"The research process of Zidong Taichu large model is also the embodiment of our mission of self-reliance and self-improvement in science and technology, which stems from Zidong Taichu's full-stack localization of basic software and hardware system. In the context of the current fierce international competition, this provides a solid foundation for the independent and controllable artificial intelligence industry in the mainland. At the report meeting, Wang Jinqiao's words were sonorous and powerful.
Wang Jinqiao, president of Wuhan Institute of Artificial Intelligence, gave a speech at the report meeting. Photo by Yangtze River reporter Peng Nian
Recall that in the summer of 2022, Wuzhiyuan was established in a temporary office in Optics Valley Bio-City, with a start-up team of less than 50 people. After two years of development, the office area of Wuzhi Institute has increased by 7 times, the R&D team has grown to 160 people, a multi-modal artificial intelligence industry consortium has been established, more than 100 member units have been absorbed, and 10 artificial intelligence start-ups have been introduced and incubated. Team leader Wang Jinqiao was also awarded the title of "Wuhan Model".
Smart Manufacturing, Smart City, Smart Cultural Tourism, Smart Healthcare, Smart Education...... With the policy support of the state, provinces and cities and the East Lake High-tech Zone, a number of major scientific and technological achievements of Wuzhi Institute will be transformed and implemented one after another.
"As the core agglomeration area for the development of Wuhan's new generation of artificial intelligence industry, Optics Valley actively seizes the opportunity of Wuhan to build a science and technology innovation center with national influence, and is committed to building a 'World Optics Valley' with global influence." Wang Jinqiao believes that Optics Valley has initially built a complete chain of artificial intelligence industry, and has formed a number of domestic leading core technologies and characteristic products with independent intellectual property rights in the fields of intelligent robots, intelligent networked vehicles, machine vision, etc., and Wuzhi Institute has the ability to consolidate the computing power foundation of the digital economy for Optics Valley. As the artificial intelligence industry gathers enterprises in Optics Valley, a large-scale model innovation and application ecology with Wuzhi Institute as the center has been formed, and a series of artificial intelligence solutions are also moving from Optics Valley to the whole country.
(Editor-in-Chief on duty: Chen Zhiyuan, Duty Director: Wang Chong, Editor-in-Charge, Li Zhiheng)