Tencent Intelligent Mobility VP: Let AI Become the "Water and Electricity" of the Automotive Industry

A few days ago, Zhong Xuedan, vice president of Tencent Intelligent Travel, accepted an interview with Interface News, which analyzed the situation of car companies in the field of AI, and also shared some of his views on AI and travel. In Zhong Xuedan's view, high-performance arithmetic, accurate data closed loop, rapid iteration of the AI model, is the three major underlying elements of automotive intelligence, but it is also the weak link of many car companies. This also leads to a series of challenges faced by car companies in the process of transforming AI technology into AI applications: training models requires a large amount of high-quality data, car companies have a huge amount of complicated data but do not know how to filter and refine it for use in a large model, and are faced with the "scale-efficiency paradox". In terms of modeling capability, the layout of infrastructure and tool chain of car enterprises is not comprehensive enough, and the Scaling Law based on pre-training has also reached a bottleneck; when AI applications are really landed on the car side, it is difficult for car enterprises to find the best way to transform. Combined with the above, Zhong Xuedan analyzed Tencent's current travel positioning: the digital assistant and ecological co-builder of the automotive industry. He said that Tencent's advantage lies in its unique data closed-loop capability, rich experience in model training, and extensive user connection capability, which can transform AI capability into the realization of user needs. The first two determine the ability, effect and quality of the model, and the latter is related to the user's core experience. Finally, Zhong Xuedan also said that the automotive industry has entered the stage of deep and comprehensive application of AI models. Zhong Xuedan also predicted the "real 'qualitative change' moment of the automotive intelligence tide": when we solve the problems of data engineering, model optimization and application scenario access, we will ultimately let the automotive industry use AI tools and make better AI applications, just like water and electricity.