Jensen Huang discusses three major breakthroughs in AI models over the past year: open-source models, physical AI, and agent-based AI.

Nvidia CEO Jensen Huang discussed three major breakthroughs in AI models over the past year at the Davos Forum. “Three major things happened at the AI ​​model level last year. First, when these models first appeared, there were many illusions, but last year, these models could be applied to research, reasoning, planning, and answering questions without prior training in the relevant fields—this led to Agentic AI,” Huang said. He added that the second major breakthrough came from open-source models. The launch of DeepSeek, the first open-source inference model, was a significant event for most industries and companies. Since then, the open-source inference model ecosystem has flourished, with many companies, research institutions, and educators utilizing open-source models for various purposes. Huang stated that the third area of ​​significant progress was in physics AI. Physics AI can not only understand language but also the physical world, such as understanding biological proteins, chemistry, and physics. In the field of physics, AI can understand fluid dynamics, particle physics, and quantum physics. (Yicai) (Jinshi Data APP)