In the field of AI smartphones, OPPO is conducting post training on Qianwen based on the PAI platform to support different AI scenarios; The visual Q&A end-to-end solution created by Honor and Alibaba Cloud has increased the recognition rate of image segmentation scenes by nearly 40% and reduced latency by 30%
Recently, the "Action Plan for Accelerating the High Level Application of Artificial Intelligence in All Areas, All Times, and All Industries in Guangdong Province" was released, focusing on seven major directions including "artificial intelligence+" scientific research, agriculture and rural areas, traditional industries, emerging industries, service industries, governance capabilities, and people's well-being, to promote the empowerment of artificial intelligence in thousands of industries and households.
Dr. Lu Ka, a children's AI terminal brand, recently announced that its AI learning camera has been fully integrated into the Qianwen model. As of now, the cumulative user base of this product has exceeded 250000, with an average monthly photo recognition of over 50 million times, and is expected to cover more than 20 countries and regions worldwide. The brand also defines the AI learning camera as a "multimodal AI companion terminal for children". After connecting to Qianwen, the AI learning camera has achieved functions such as image understanding, visual inference, OCR recognition, and real-time interaction, and has reached SOTA level in multiple evaluations, with end-to-end latency controlled within 1 to 2 seconds.
Not only Dr. Lu Ka, but recently, Zhuji Power has released the latest generation of humanoid robot LimX Luna. This product is built on Alibaba Cloud's full stack AI and integrated with the Qianwen Omni model. The training of embodied intelligence "brain" models and operational control "cerebellum" models requires massive amounts of multimodal data, with extremely high requirements for training throughput and storage. At present, the scale of training data for each interval power has reached 1.5PB, and the instantaneous throughput during training peaks can reach hundreds of TB.
Chen Liwei, General Manager of Solutions for the Southern Region of Alibaba Cloud Intelligence Group's Public Cloud Business Unit, stated that over 150000 smart hardware manufacturers have integrated into the Qianwen big model, and top domestic embodied intelligent enterprises and global TOP10 mobile phone manufacturers have also integrated into Qianwen. In addition, Qianwen also covers almost all scenarios such as consumer level, industrial level, travel, home, security, etc. The continuous upgrade of these physical AI carriers is supported by Alibaba Cloud's full stack AI technology.
Wu Chenye, Assistant Dean of Polytechnic of the Chinese University of Hong Kong (Shenzhen) and Assistant Dean of Hetao College of Shenzhen, believes that "Internet plus" is more about connecting hardware to the Internet, adding an App portal, or realizing remote control. But the big model does not bring simple networking, but rather enables hardware to have certain understanding and collaboration capabilities. After hardware integration into the big model, it is no longer just a passive execution of instructions, but can become a "buddy" in work, study, and life. For example, learning machines are not just for problem solving, robots are not just for performing actions, and AI glasses are not just for taking photos and displaying information, but can participate in human thinking, expression, and action processes.
It is reported that by 2025, the scale of Guangdong's core artificial intelligence industry will exceed 300 billion yuan, with a year-on-year growth of over 40%, accounting for about 1/4 of the national scale. The computing power chips, industry models, intelligent terminals, and application empowerment are all at the forefront of the country.
In the future, Guangdong will innovate and optimize policy implementation around the actual needs of enterprises, and allocate more resources such as platforms, projects, talents, and funds to and from enterprises, supporting enterprises to become the main body of artificial intelligence technology innovation, product creation, and application landing. We will focus on cultivating independent ecological leading enterprises, support chain leading enterprises to connect upstream and downstream industries, and link large, medium, and small enterprises, striving to become all-round champions of "intelligent design+scale manufacturing". (Looking into the New Era)
Recommended Reading Change it
