Demis Hassabis, best-known for his work with DeepMind, is one of three recipients of the Nobel Prize for Chemistry.
Professor Demis Hassabis, a co-founder of DeepMind who began his programming career working for Bullfrog and Lionhead, has been awarded the Nobel Prize for Chemistry, alongside Professor John Jumper and Professor David Baker. Hassabis and Jumper’s share of the award is for their “complete revolution” in the prediction of protein structures through the AI tool AlphaFold2.
AlphaFold first became available to researchers in 2018, and 2020’s AlphaFold2 is the second iteration (the third, AlphaFold3, was announced in May this year). It is an AI dedicated to predicting protein structures, created by DeepMind, which has now been used to predict the structure of almost all known proteins, a feat unimaginable a mere decade ago, as well as create new proteins by David Baker and his researchers.
Per the Nobel citation, Professor Baker receives half of the Nobel “for computational protein design”, while the other half is shared jointly between Professor Hassabis and Professor Jumper of DeepMind “for protein structure prediction”. The winners receive a share of 11m Swedish kronor ($1.05 million).
“One of the discoveries being recognised this year concerns the construction of spectacular proteins,”says Heiner Linke, chair of the Nobel Committee for Chemistry. “The other is about fulfilling a 50-year-old dream: predicting protein structures from their amino acid sequences. Both of these discoveries open up vast possibilities.”
The Nobel press release says AlphaFold2 has been used “to predict the structure of virtually all the 200 million proteins that researchers have identified.”
“Since their breakthrough, AlphaFold2 has been used by more than two million people from 190 countries,” continues the org. “Among a myriad of scientific applications, researchers can now better understand antibiotic resistance and create images of enzymes that can decompose plastic.
“Life could not exist without proteins. That we can now predict protein structures and design our own proteins confers the greatest benefit to humankind.”
The list of actual and potential medical benefits from this work is enormous: Proteins are commonly described as the “building blocks” of all human body cells. AlphaFold 2’s work is used in countering antibiotic resistance and lies behind the creation of plastic-guzzling enzymes.
“It’s an incredible honour, you know, it’s the big one really,” laughs Hassbis on a phone call shortly afterwards. “My mind went blank it was so incredible, an unbelievable experience. I had a whole normal day of work ahead of me but I guess that has to change now.”
Hassabis has previously paid credit to his early days, encouraging children to not only be interested in games but in how they’re created, and that this is what ultimately put him on the path of his AI passion.
“I think that started sparking off in my mind ideas about how does the chess computer play chess and learning about that,” Hassabis said in 2020. “Many children start by playing games, like I did, and then getting into programming and then using this incredible tool, the computer, to create things.”
Hassabis worked for Bullfrog before attending university, and after graduation began working at Lionhead, where he was lead AI programmer on Black & White. Thanks to the game’s protracted development it would release in 2001, by which time Hassabis had left to found Elixir Studios, where he would act as executive director on Republic: The Revolution and Evil Genius.
He would then move more towards academia, completing a PhD and working at several universities, before in 2010 co-founding the AI company DeepMind. The goal from the start was to create a general-purpose AI that can be used for anything, but it began with videogames from the ’70s and ’80s. In the early days DeepMind’s AI’s trained on games like Breakout and Pong, learning the rules in order to master the game, before the company would focus on more complex games like Go and even Starcraft 2.
DeepMind was acquired by Google in 2014. In around 2016, with all that videogame training behind it, the company’s focus turned to protein folding and AlphaFold: Software which, almost from the start, has made stunning advance after “stunning advance” in protein folding.
About Post Author
