On the eve of CES, in a conversation between Fei-Fei Li and Lisa Su, CEO of MD, she emphasized that the world's potential for AI is actually underestimated.

Language models have made tremendous progress in the past few years, but intelligence is not limited to language. For humans, we rely on spatial intelligence to act, judge, and create in the real world. What truly excites her is that generative AI is entering a new phase—machines are not only able to understand information, but are also beginning to generate and explore the three-dimensional world while adhering to physical laws.

In her description, creators can shape spaces as if sketching in the real world, architects can "walk into" the design before construction begins, and intelligent agents can learn in a highly realistic digital world before entering reality, thus becoming safer and more efficient. She believes this is no longer just a vision of the future, but the starting point for AI to enter the next chapter.