Edited by Search Map East (*chedongxi).
Author |Mingcheng.
Edit | juice
Che Dongxi reported on December 22 that a few days ago, the "AI+ Travel" industry case salon jointly sponsored by Guanghua School of Management of Peking University and Tencent Cloud was held at Peking University. At the salon, four Peking University people and experts in the field of smart cars shared their understanding of how AI-driven cars are new intelligence.
Zhong Xuedan, vice president of Tencent Smart Mobility, pointed out that the combination of large models and cars has a broad application space and scenarios, and the new intelligence of automobiles covers the cockpit field and autonomous driving technology, and the voice interaction rule mode has changed to natural interaction and multimodal interaction. The application of large models is of great significance for data annotation, autonomous driving R&D, and cloud AI training.
Gou Xiaofei, vice president of Li Auto Intelligent Space, pointed out that voice interaction is a more suitable form of interaction for cars, and the large model will help the development of voice interaction. However, conversational interactions are still challenged by open conversations and predefined user interfaces.
Wang Yongliang, deputy general manager and CTO of Wutong Car Link Technology, mentioned that Wutong has made some explorations in the large model, and the self-developed language model starts from reconstruction, has the ability of continuous dialogue, and has the characteristics of theory, integration and vertical domain. In addition, he also mentioned that the application of large models in voice interaction, graphic recognition, intelligent driving, etc., has brought more capabilities to the intelligent cockpit, which can realize functions such as comprehensive intention understanding and AI automatic adaptive vehicle control.
Wang Tiemin, associate professor of the Department of Organization and Strategic Management at Peking University's Guanghua School of Management and co-director of the Peking University Management Case Study Center, shared the role of AI in the intelligent automotive industry from the perspective of business schools. She pointed out that in the era of large models, cooperation in the automotive industry is necessary and complex. The School of Business has noticed the new demand for management talents in the technology industry, and has enhanced their understanding of business management through special industry courses and promoting the exchange between science and technology innovation and management.
In general, the experts at this salon elaborated on the development direction, opportunities and challenges brought by the AI model to the intelligent automotive industry, which will profoundly empower the entire automotive industry chain and bring a profound industry change. In this change, scientific and technological talents need to understand management in order to better promote technological innovation and implementation in cooperation.
Zhong Xuedan, vice president of Tencent Smart Mobility, pointed out that the automotive industry continues to make efforts in electrification and intelligence, and the important dimension is intelligence, which is very important for enterprises and the entire industry.
Edit the search image.
Zhong Xuedan, vice president of Tencent Smart Mobility.
First of all, the application of large models in voice, multi-task, and multi-modal interaction will strengthen the human-computer interaction mode, and at the same time improve the production and output of cockpit content.
In terms of intelligent driving, the large model will enable the iteration of technology and the integration of perception decision-making, and at the same time will improve the user's perception experience of driving. In addition, the large model will also play a key role in driving change in the cloud through the optimization of multiple capabilities such as computing power, real-time mapping, and **.
The combination of large models and the automotive industry has a wide range of applications in R&D, manufacturing, marketing, and service, and may become the basis for the integration of infrastructure and infrastructure, bringing about changes to the entire industry and industry.
Gou Xiaofei, Vice President of Li Auto's Intelligent Space, shared the topic of AI-based human-computer interaction. He pointed out that the evolutionary history of human-computer interaction in the past closely matches the evolutionary history of automobiles and consumer electronics. As a multi-person service space terminal, the traditional touch interaction is not suitable in the car because of its distance limitations. In the process of driving, the driver's touch also needs hand-eye interaction, and the driver's attention is mainly focused on the road.
Edit the search image.
Gou Xiaofei, Vice President of Intelligent Space of Li Auto.
For many years, language has been a form of assisted interaction in automobiles because of the lack of semantic comprehension before. However, the emergence of large models has improved the ability to understand language, and conversational interaction is expected to replace touch interaction as the mainstream.
In order to achieve conversational interaction, the industry faces two core problems: one is that open dialogue needs to match open output, that is, all UI will be automatically generated according to the language logic;Second, in the dialogue interaction, the form of the application may become a virtual assistant, integrating various capabilities and providing complete logic.
Wang Yongliang, deputy general manager and CTO of Wutong Car Link Technology, shared the application and practice of the industry model in the field of intelligent cockpit, they defined intelligence as an intelligent cockpit by building the intergenerational development concept of theory-technology-product, referring to the experience of intelligent driving classification, and Wutong Automobile began to build intelligence in 2020, and now it has reached the strategic brain 3Stage 0.
Edit the search image.
Wang Yongliang, deputy general manager and CTO of Wutong Auto Link Technology.
The strategic brain of Wutong Automobile has gone through different stages of development, from the implementation of simple rules to the realization of complex scene orchestration and multi-scene linkage, and finally reached the level of L3.
The reason why Wutong Automobile has been able to achieve these developments is due to the large-scale model theoretical base provided by Tencent on the one hand, and the practical opportunity to cooperate with Changan on the other hand.
Wutong Automobile has explored in terms of large models, and their self-developed language model is based on Tencent's hundreds of billions of parameters, and has added support in some fields, and has now reached trillions of parameters, with continuous dialogue and intelligent voice interaction capabilities. The characteristics of the large model of Wutong Automobile include theory, integration and vertical domain, and they pursue the concept of device-cloud integration, focusing on efficiency, stability and safety.
Wang Tiemin, associate professor of the Department of Organization and Strategic Management at Guanghua School of Management at Peking University and co-director of the Management Case Study Center of Peking University, pointed out that the application of AI in the industry is an innovation at the level of technology paradigm, which is not only the development of technology, but also involves factors such as economy and system. The innovation of the technological paradigm needs to take into account both demand and supply factors.
Edit the search image.
Wang Tiemin, associate professor of the Department of Organization and Strategic Management, Guanghua School of Management, Peking University, and co-director of the Peking University Management Case Study Center.
Smart travel includes multiple elements such as cloud, vehicle, road, network, and map, among which the vehicle end is an important entrance. The application of artificial intelligence in the vehicle can change the manufacturing, marketing, and after-sales service of the automotive industry, improving user experience and dealer efficiency. The application of AI technology can bring about improvements.
The application of new technologies will lead to the reshaping of industrial boundaries and the reconstruction of competition and cooperation relationships, especially in the automotive industry. The relationship between technology companies and the vehicle manufacturing industry needs to be redefined, but it won't be easy. Traditional automotive companies may be worried about being controlled by software, so they need to find a balance between soul and body.
Over time, the automotive industry will see more cross-border cooperation and specialization, rather than all companies taking the path of full-stack self-development. For enterprises that are not full-stack self-developed, cross-industry integration is necessary. Experience management is also very important in the implementation of AI to promote the new intelligence of automobiles.
The School of Business has noticed the new demand for management talents through science and technology, and has enhanced their understanding of business management through specialized industry courses and promoting the exchange between science and technology innovation and management. The School of Business attaches great importance to the ability to create knowledge, and promotes the creation and understanding of knowledge by combining software and hardware knowledge through the exchange and expression of experience.
Professor Wang Tiemin also put forward the view that scientific and technological talents should also understand management in the AI era, which shows the deep impact of large models on the entire automotive industry.
In the future, large models will have a deeper impact on the automotive industry, and will promote the implementation of many technologies such as end-to-end autonomous driving, natural language interaction, and intelligent cockpits. Rapidly changing technologies and industries need more management talents who understand technology, and technical workers also need to understand management better, so that more innovative achievements can emerge.