2 1 Shi Ji Jing Ji Bao Dao·2025-04-03 02:12Group 1: Automotive Industry - Xiaomi's customer service clarified that their intelligent driving system is only an auxiliary driving feature, requiring human intervention when necessary. The system includes a fatigue monitoring camera that alerts the driver if fatigue is detected, and if no action is taken, the vehicle will gradually slow down until it stops [1] - Li Xiang, founder of Li Auto, called for a unified standard for the Chinese terminology of autonomous driving, suggesting clearer definitions for different levels of driving assistance to avoid user confusion [1] - CATL responded to inquiries regarding the battery used in a Xiaomi SU7 vehicle that caught fire, stating that the battery involved was not produced by them [2] Group 2: Technology and AI - Microsoft's CTO predicted significant breakthroughs in AI agents' memory capabilities within the next year, which will change their interaction with humans [3] - Meta's AI research head Joelle Pineau announced her departure from the company, effective May 30 [3] - Alibaba's Tongyi Qianwen model topped the global open-source model rankings, with over 200 models released and more than 100,000 derivatives, surpassing the Llama series from the US [4] Group 3: Corporate Developments - Apple CEO Tim Cook sold company stock for over $24 million, part of a restricted stock unit plan from 2020, with other executives also selling shares [2] - Microsoft reportedly closed its AI and IoT lab in Shanghai, which had supported numerous innovation projects and trained nearly 10,000 professionals since its establishment in 2019 [5] Group 4: Investment and Financing - Yuan Ding Intelligent completed nearly 1 billion RMB in strategic financing, with Fluidra as a major investor, enhancing its global market reach [6] - Pudong Venture Capital Group participated in a multi-hundred million RMB Pre-IPO investment in DeepBlue Technology, which focuses on AI applications across various sectors [6] Group 5: New Product Releases - Baidu announced the official release of the PaddlePaddle 3.0 framework, featuring significant technological innovations to enhance model training and inference efficiency [7] - DeepSeek's new patent aims to reduce network resource consumption during data collection, improving data quality and efficiency [7]