何小鹏：物理AI或将在未来三年迎关键突破

Group 1 - The core observation is that significant breakthroughs in AI over the next three years are likely to occur in the physical domain, such as autonomous driving and robotics, rather than in the digital realm [1] - The entrepreneurial atmosphere in the AI sector is particularly vibrant in the U.S., especially in the SaaS and physical AI robotics sectors, with high valuations and concentrated startup projects [1] - There is a notable divergence in the approach to AI robotics between China and the U.S.; Chinese companies tend to focus on hardware technologies like joints and controls, while U.S. companies are more inclined towards model-level innovations [1] Group 2 - There is no current AI bubble, as the industry is still in its early stages of driving social change, with China exhibiting more rational valuations focused on application, while the U.S. has higher valuations centered on frontier exploration [2] - The development of AGI (Artificial General Intelligence) is still in progress, with current AI capabilities primarily based on imitation and reinforcement learning, lacking true creativity [2] - Achieving true AGI will require breakthroughs in multi-modal understanding, world model construction, continuous learning, and long-term planning, which may take several years and depend on further advancements in underlying technologies [2]