Just when they believed they were passed over for this year's Nobel Prize in chemistry, two scientists from Google's DeepMind AI research team got the call -- mere minutes before they were announced as honorees.
Demis Hassabis, the CEO of Google's DeepMind, and John Jumper, the project's American director, shared the prize for their work on AlphaFold2, an AI model that can predict protein structures. The two were co-honorees with David Baker, a University of Washington scientist who has been using amino acids and computational power to create new kinds of proteins.
Hassabis and Jumper both said they received word from the Swedish prize organization just before the news went out; emergency phone calls and texts ultimately reached Hassabis' wife and another member of the DeepMind team. "We got the call very late. We were assuming it wasn't happening," Hassabis said in a press conference held by Google after the Wednesday announcement. "I tried to sleep in," Jumper added. "I couldn't sleep last night."
The AlphaFold project was first presented in 2020 and has since predicted the structure of 200 million proteins identified by researchers. AlphaFold2, for which Hassabis and Jumper won the award, has been used by more than 2 million people in 190 countries. In the press conference, the two said that a version in the works, AlphaFold3, will be released to the scientific community for free.
This year's Nobel Prize for Physics, awarded one day earlier, also recognized pioneering work in AI, which revealed "a completely new way for us to use computers." Geoffrey Hinton, of the University of Toronto, and John Hopfield, of Princeton University, shared the prize for using physics to train neural networks -- systems inspired by the workings of the human brain -- and thus enabling the machine learning that drives much of what artificial intelligence can accomplish.
Hinton, known as the "godfather of AI," worked for a time at Google, but left in 2023 citing concerns about the risks that artificial intelligence poses. On Tuesday, he noted both the positive implications, such as advances in health care, and the negatives and the sheer unknowables as AI rapidly evolves. "We have no experience of what it's like to have things smarter than us," he said, as reported by The New York Times.
"AI as the ultimate tool"
The Nobel committee called AlphaFold2 a "stunning breakthrough." In the press conference, Hassabis and Jumper acknowledged that their work is only the beginning of AI-assisted technology that could speed up the development of medical treatments from years to months and that will help researchers understand what Hassabis called "fundamental mechanisms in biology."
"I kind of see AI as potentially the ultimate tool for accelerating science and scientific knowledge," Hassabis said.
Hassabis and Jumper will split the prize of 11 million Swedish kronor (about $1.06 million) with Baker.
The two credited the team at Google and many other scientists who created the foundational work that their research built upon. "It's humbling," Jumper said. "Every time we train AI, every data point is years of effort from someone training to be a Ph.D. student or someone who's already gotten their Ph.D. ... Every day it's wonderful to see the work that the scientific community has done on top of AlphaFold and I can't wait to see the next breakthroughs."
While AI was a significant part of AlphaFold, instrumental in identifying patterns that humans wouldn't be able to find, Hassabis pointed out that a lot of human work went into the project. "It wasn't just 'AI did this,'" he said. "It was an iterative process. We developed, we did research, we tried to find the right combinations between what the community understood about proteins and how we build those intuitions into our architecture."
"AI was the toolbox in which we got to this incredible work," Hassabis said