Huawei and Zhuoyu Executives Debate Whether Intelligent Driving Must Go Through L3 Stage

On April 11, at the High-Level Forum on Intelligent Electric Vehicles, a disagreement emerged between Yu-Zhi Jin, CEO of Huawei's Intelligent Automotive Solution Business Unit (BU), and Shao-Jie Shen, CEO of Zhuoyu Technology, regarding the evolutionary path of autonomous driving technologies. Jin argued that Level 3 (L3) autonomy is an indispensable stage that cannot be skipped. He emphasized that Level 4 (L4) autonomous driving must achieve a safety level ten times higher than human driving and that this can only be accomplished by accumulating real-world data and helping users adapt through L3 deployment. He further stressed that L3 marks a critical transition in shifting accident liability from vehicle owners to manufacturers, which is essential for establishing compliant regulatory and insurance frameworks. Jin disclosed that, as of March 31, 2026, Huawei’s Qiankun intelligent driving active safety system had cumulatively prevented 4.79 million potential collisions, with total assisted-driving mileage exceeding 9.5 billion kilometers. He noted that pure assisted driving is already 4.2 times safer than human driving and predicted that 2026 will be the "Year One" of global autonomous driving. In contrast, Shen contended that it is feasible to leap directly from Level 2 (L2) to L4, citing ambiguous liability allocation as a key practical drawback of L3. He asserted that in the era of large models, L2, L3, and L4 can all evolve from a common technological foundation, and advancing directly to L4 would help clarify responsibility. Introducing the concept of "Mobile Physical AI," Shen described intelligent driving as merely the initial form of physical AI. He revealed that Zhuoyu is developing a native multimodal foundation model, scheduled for deployment in passenger cars and heavy-duty trucks within 2026, and that its end-to-end 4.0 system already possesses defensive driving capabilities.