As part of the series of activities of the Shanghai Automobile Culture Festival, on April 17, the 2nd Automotive Artificial Intelligence Conference of Gasgoo 2024 opened in Shanghai International Automobile City. The conference focused on hot topics such as large-scale model technology, automotive robots, and AI intelligent cabins, and discussed how AI can reshape the automotive industry.
At present, large-scale model technology is leading the automotive industry to a new height with its powerful intelligent driving force. Specifically, how does the large model reshape the pattern of the automotive industry? He Weimin, deputy general manager of iFLYTEK's intelligent vehicle division, believes that the launch of ChatGPT in 2022 is the fourth wave in the history of artificial intelligence, and the large model will bring great changes. Taking the iFLYTEK Xinghuo cognitive model as an example, in 2023, iFLYTEK will release the independent and controllable iFLYTEK Xinghuo cognitive model, and actively promote the application of the large model in multiple industries. For car companies, iFLYTEK has launched a digital and intelligent global solution from hardware, large model base to scenario requirements and applications, deeply integrating large models with the automotive industry, thereby leading new changes in the automotive industry.
"This solution is to build an enterprise innovation and development platform by integrating the all-in-one machine of the intelligent computing platform with the Spark large model and the knowledge platform. Based on this platform, engineers in various fields of automotive companies can quickly develop applications for different scenarios and solve some problems they encounter in practical work. He Weimin said that iFLYTEK's intelligent vehicle business segment covers five aspects, including intelligent sound, intelligent cockpit, and intelligent driving, and realizes the all-round improvement and empowerment of artificial intelligence on the interactive experience of automobiles.
Multimodal capability is one of the seven core capabilities of the iFLYTEK Xinghuo cognitive model, and Weimin said that in the automotive field, through the combination of voice, vision and the core technology of the large model, it can create a natural, free and intelligent interactive experience, bring interactive changes to the vehicle, integrate the voice recognition of visual detection, make it clearer, improve the overall noise reduction effect, and the application of eye interaction and gesture interaction makes the interaction between people and vehicles more intelligent and diverse.
In the shaping of intelligent experience, the concept of "AI native" has attracted more and more attention. "AI Native" deeply integrates AI technology with automotive products, making the products themselves intelligent and adaptive. Pan Yunpeng, vice president of Jiyue Automotive's intelligent software, shared "how AI natively reconstructs the intelligent experience and how it is practiced on the Jiyue 01 car". Pan Yunpeng believes that "software-defined vehicles" are gradually transforming into "AI-defined vehicles", and for the future development of automobiles, it is necessary to reconstruct the design of the entire intelligent car with "AI native thinking".
"Based on this understanding, when we designed the company's first product, we believed that as an automotive robot, the Jiyue 01 needed to have three core capabilities: natural communication, free movement, and self-growth. We need to enable smart cars to think and act like people, and to evolve. Pan Yunpeng said that the company's goal is to combine Baidu's AI artificial intelligence and autonomous driving in the past ten years to make intelligent fault products. Jiyue 01 has integrated AI large model technical capabilities such as Wenxin Yiyan and Baidu Map.
The rise of AI has also injected impetus into the development of full-stack solutions for intelligent vehicles. Meng Chao, CTO of ZeroBeam Technology Co., Ltd., believes that now in the second half of "software + AI to define cars", it is necessary to establish a new data closed-loop infrastructure based on users, scenarios, products and ecological big data, and combine the vertical scenario implementation of AI large models in the automotive industry to realize the continuous self-learning, self-evolution and self-growth of smart cars.
"ZeroBeam has always provided intelligent full-stack solutions for automotive intelligence, and has released full-stack 3.0 solutions. The solution is pre-installed with an AI operating system, which will realize the integration of intelligent driving, intelligent cockpit, and the entire central computing and intelligent interconnection, with self-learning, self-evolution and self-growth capabilities. Meng Chao said that the application of artificial intelligence in the automotive vertical field is a long road, and he hopes that relevant practitioners in the industry will work together to build an artificial intelligence ecology and create a cost-effective and excellent experience of intelligent car products.
"We gathered 12 car companies, 12 technology companies, and 19 exhibitors to display and communicate on site. Zhou Xiaoying, CEO of Gasgoo, said that he hopes to create a good platform for everyone to communicate in an all-round way, better understand the current situation and future trend of China's intelligent vehicle development, and help the intelligent development of the automotive industry and the construction of a good ecosystem.
Written by: Li Pin
Cinematographer: Huang Jiaxing
Editor: Yang Yujie