字节跳动否认造车 豆包大模型、火山引擎已与多家车企合作

Core Viewpoint - ByteDance is reportedly not pursuing any plans to manufacture cars, despite rumors suggesting otherwise. The company emphasizes its focus on providing AI solutions for the automotive industry through its Volcano Engine platform [1][2]. Group 1: ByteDance's Automotive Strategy - ByteDance's Volcano Engine is actively collaborating with numerous automotive clients, providing an efficient algorithm training platform, but it does not have plans for autonomous driving business [2]. - Over 90% of mainstream automotive companies are partnering with Volcano Engine for AI innovations, covering various scenarios from smart cockpits to intelligent marketing and autonomous driving labeling [2]. Group 2: Collaborations and Innovations - The SAIC Roewe M7 DMH, launched in September 2022, features the Doubao Deep Thinking Model, enabling the vehicle to interpret vague user commands and understand true intentions through long-term memory capabilities [4]. - BYD has integrated the Doubao model with its DiLink system in just over four months, enhancing voice interaction, content recommendations, and travel services [4]. - Xiaomi is collaborating with Volcano Engine on multiple fronts, including stable foundational computing facilities and AI cloud-native acceleration solutions, which have significantly improved training stability and simulation efficiency [4]. Group 3: Recent Automotive Showcases - At the Guangzhou Auto Show in November, several models from brands like Mercedes-Benz, SAIC Audi, and Changan Mazda showcased their collaboration with Volcano Engine, featuring advanced human-machine interaction systems [5]. - The Mercedes-Benz electric CLA model, equipped with a new intelligent interaction system, has improved interaction efficiency by 50%, with a wake-up speed of 0.2 seconds and an end-to-end execution speed of just 0.8 seconds [5]. - The SAIC Audi E5 Sportback and Changan Mazda EZ60 also demonstrated enhanced capabilities in understanding user intent and executing commands through voice instructions, showcasing the integration of the Doubao model [5].