When John Jumper joined Alphabet-owned, London-based AI lab DeepMind in 2017, he was tasked with building algorithms to predict the 3-D shape of proteins, which is key to developing an array of new drugs—and a job that has long daunted mere humans. AlphaFold, the software his team he developed, went on to win a biannual protein-prediction competition called CASP last November, dazzling the scientific community and inspiring comparisons to such breakthroughs as the 1953 discovery of DNA’s double-helix structure. “I always believed it should be solvable,” says Jumper, who has a background in physics, chemistry, and mathematics. He credits DeepMind’s success to both concerted teamwork and a series of epiphanies along the way about how proteins work. In July , the company followed up on its achievement by publishing the scientific paper and source code behind AlphaFold, and then releasing predictions for all the human body’s proteins as well as those for 20 other organisms.
AlphaFold’s real impact will come as researchers at DeepMind and elsewhere leverage its insights. When future pandemics strike, it could speed the creation of vaccines, for example. But Jumper is particularly excited by the potential to help treat rarer, under-researched ailments such as tropical diseases. And he stresses that AlphaFold will lead to additional breakthroughs: “That this problem that was intractable for 50 years has been solved is really promising for what [AI-based medical research] is going to look like in five and 10 years.”
Read more about Fast Company’s Most Creative People in Business 2021