Former DeepMind Scientist Yuan Cao: AI Winning Nobel Prize Independently Will Take Another Two to Three Decades

According to the AI Watch monitoring, Cao Yuan, former Senior Research Scientist at Google DeepMind and Co-Founder/CEO of Unreasonable Labs, believes that AI for Science is entering a period of explosion. However, for AI to truly make Nobel Prize-level scientific discoveries independently, it will still need another two to three decades. Cao Yuan has previously been involved in projects like Gemini, and his current entrepreneurial direction is also focused on enabling AI to discover new knowledge.

He believes that the current biggest bottleneck is validation. Writing code can be tested immediately, and mathematics can be gradually checked and proven using Lean. However, biology, materials, and physics all ultimately require real experiments. A single experiment may be costly or may take a long time, making it difficult for AI to iterate rapidly and continuously like a Coding Agent.

Even more challenging is "creating new concepts." Cao Yuan believes that current models are good at searching for answers within existing knowledge, definitions, and theorems, but they are not yet capable of abstracting new mathematical objects, definitions, and theories like top scientists. He refers to this "concept abstraction" as the "last mile" of AGI.