谷歌深度思维首席执行官称：中国人工智能模型与美国的差距仅为“数月”

Core Insights - The CEO of Google DeepMind, Demis Hassabis, stated that the gap between Chinese AI models and those of the US and Western countries may be only "months" [1][10] - Hassabis noted that while Chinese companies have shown the ability to catch up technologically, they have not yet demonstrated the capacity for breakthrough innovations in AI [2][11] Group 1: Current State of AI Technology - Hassabis emphasized that the performance of Chinese AI models is closer to that of the US and Western technologies than previously anticipated, suggesting a narrowing gap [1][10] - He mentioned that approximately one year ago, the Chinese AI lab DeepSeek launched a model that caused a significant market impact, despite being built on relatively outdated chips [1][10] Group 2: Challenges and Limitations - Despite advancements, Hassabis pointed out that Chinese companies have not proven their ability to achieve disruptive innovations, such as developing breakthrough technologies like new transformers [4][13] - The CEO of Alibaba's Tongyi Qianwen team indicated that the probability of Chinese companies surpassing US tech giants in AI within the next three to five years is less than 20% due to the significant scale of US computational infrastructure [6][15] Group 3: Industry Perspectives - Other tech leaders, such as NVIDIA's CEO Jensen Huang, acknowledged that the US is not far ahead in the AI race, stating that both countries are relatively equal in AI models and infrastructure [5][14] - Analysts have expressed concerns that the inability of Chinese companies to access advanced NVIDIA chips could widen the technological gap between the US and China in AI [14] Group 4: Innovation and Mindset - Hassabis attributed the lack of frontier technology breakthroughs in China to "mindset" rather than technical capabilities, suggesting that true scientific innovation is significantly more challenging than mere imitation [7][16] - He compared Google DeepMind to the "modern Bell Labs," emphasizing the importance of exploratory innovation over merely scaling existing technologies [8][17]