I am thrilled to hear that the #NobelPrize in Physics has been awarded for work in AI and neural networks. Caltech has been the birthplace of AI. Many are not aware that the famous NeurIPS conference was started at Caltech and had its genesis from a course that John Hopfield, Richard Feynman, and Carver Mead taught, โThe Physics of Computation.โ
When I arrived at Caltech in 2017, I wanted AI to touch the very foundations of scientific discovery. AI at that point was mostly focused on image and text domains, and most domain scientists were skeptical about how useful AI would be for them. I co-founded the AI+Science initiative to close this gap, get researchers to collaborate across disciplines, and explore how AI could fundamentally transform science.
This led us to invent Neural Operators : an AI method that can learn physical phenomena at multiple scales. We used it to train the first high-resolution AI-based weather model, which is tens of thousands of times faster than traditional weather models. We have modeled plasma evolution in nuclear fusion and carbon sequestration in underground reservoirs. We have invented a better medical catheter that reduces bacterial contamination by 100x. We are just getting started on AI+Science.
When I arrived at Caltech in 2017, I wanted AI to touch the very foundations of scientific discovery. AI at that point was mostly focused on image and text domains, and most domain scientists were skeptical about how useful AI would be for them. I co-founded the AI+Science initiative to close this gap, get researchers to collaborate across disciplines, and explore how AI could fundamentally transform science.
This led us to invent Neural Operators : an AI method that can learn physical phenomena at multiple scales. We used it to train the first high-resolution AI-based weather model, which is tens of thousands of times faster than traditional weather models. We have modeled plasma evolution in nuclear fusion and carbon sequestration in underground reservoirs. We have invented a better medical catheter that reduces bacterial contamination by 100x. We are just getting started on AI+Science.