The laureates of the 2024 Nobel Prize in Chemistry are seen on a screen (left to right) David Baker, Demis Hassabis and John Jumper, at the Royal Swedish Academy of Sciences in Stockholm, Sweden, on Oct. 9.JONATHAN NACKSTRAND/AFP/Getty Images
For the second time this week, artificial intelligence is at the heart of a Nobel Prize.
Demis Hassabis and John Jumper of Google DeepMind, an AI research company based in Britain, have been awarded one half of this year’s Nobel Prize for chemistry for creating an algorithm that can predict the structure of a protein based on its genetic code.
They are sharing the prize with David Baker, a U.S. researcher known for using computational methods to design bespoke proteins for a range of applications and then reverse engineering how to make them.
The announcement Wednesday of this year’s chemistry prize by the Royal Swedish Academy of Sciences comes one day after University of Toronto computer scientist Geoffrey Hinton and Princeton University physicist John Hopfield were awarded the Nobel Prize for physics for their contributions to the development of neural networks, self-correcting computer algorithms that have revolutionized machine learning.
Wednesday’s winners, who will share a prize sum worth 11-million Swedish kronor ($1.4-million), reflect the growing importance of neural networks and related AI methods in the quest to explore and manipulate the molecular building blocks of life.
All three winners are still in the midst of their active research careers. Dr. Baker, 62, directs the University of Washington’s Institute for Protein Design in Seattle. Dr. Hassabis, the chief executive of Google DeepMind, and Dr. Jumper, a senior research scientist there, were born in 1976 and 1985, respectively.
Calling their developments “mind-blowing,” Johan Aqvist, a member of the Nobel chemistry committee said, “It’s almost as if it’s only your imagination that sets the limit for what you can do here.”
Proteins are complex molecular components built out of amino acids that play a role in nearly all biological processes. They are key to the workings of cells and to the structures that help cells communicate and function in larger organisms.
The DNA molecule carries the instructions for making all the proteins a cell needs to survive, but predicting the three-dimensional structure of a protein based on its genetic sequence is an infamously complex problem.
In 2018, Dr. Hassabis and Dr. Jumper debuted the algorithm known as AlphaFold, which employs AI to predict protein structure. It was based on computer code that was originally used to defeat the world’s top player of the traditional board game Go.
By 2020, a retooled version called AlphaFold2 walked away with first place in a competition, where it vastly outperformed all other approaches to the problem. By then it had reached the level where its predictions were of practical use to scientists. It has now been used to predict the structures of hundreds of millions of proteins.
The work means researchers can read the genome of an organism and understand how its proteins are shaped. In the case of a harmful pathogen, such as a virus, they can then design molecules that bind to those proteins to block their action.
Dr. Hassabis has said the same general approach that allowed AlphaFold to learn how to predict protein shapes can be used to tackle a broad range of problems across the sciences.
“The way we think about artificial intelligence is as these general learning systems that are able to potentially learn about any domain,” he said in an interview with The Globe and Mail last year.
In parallel, Dr. Baker is known for coming up with new protein shapes that meet particular needs for everything from medicines to environmental sensors. By 2003 he had demonstrated the success of a computer program that could predict how to synthesize the proteins he designed.
He has since created a number of proteins that previously did not exist in nature and can perform an array of useful functions, such as blocking variants of the COVID-19 virus or detecting the presence of fentanyl.
The work continues to evolve and now also incorporates AI into its algorithms, inspired by the progress of AlphaFold and related efforts with applications for health and environmental monitoring and sustainability.
“Our new AI methods are much more powerful,” said Dr. Baker when he was reached during the chemistry prize announcement. “I’m really excited about all the ways in which protein design can now make the world a better place in health.”
Last year Dr. Hassabis and Dr. Jumper were recipients of the Canada Gairdner International Award for their work. They join a growing number of researchers who have received the Canadian prize and gone on to win a Nobel.
Canadian winners of the Nobel Prize for chemistry include Gerhard Herzberg, a scientist with the National Research Council in Ottawa, who received the prize in 1971 for work in the field of atomic and molecular spectroscopy. Fifteen years later, John Polanyi, a professor emeritus at the University of Toronto, took home the award for discoveries related to chemical kinetics. Then Michael Smith, a biochemist at the University of British Columbia, won in 1993 for his studies related to mutations in DNA.
Wednesday’s announcement rounds out the three science Nobels for this year.
On Monday, U.S. researchers Victor Ambros and Gary Ruvkun were awarded the Nobel Prize for medicine or physiology for their discovery of microRNA, tiny molecules that are crucial for gene regulation.
This year’s Nobel Prize for literature is set to be announced Thursday, and the Nobel Peace Prize winner will be named Friday. The final prize, in economics, will be announced Monday, Oct. 14.