This article covers how the 2024 Nobel Prize in Chemistry recognized breakthroughs in protein science achieved through artificial intelligence. It discusses how AI tools like AlphaFold and Rosetta revolutionize our understanding and engineering of proteins, opening new possibilities in medicine, environmental science, and more.
- AI in Protein Folding: Hassabis and Jumper used AI to predict protein structures, solving a 50-year-old challenge in chemistry.
- Revolutionizing Research: AlphaFold significantly reduced the time needed to determine protein structures, accelerating scientific discovery.
- Designing Novel Proteins: David Baker developed methods to create entirely new proteins, expanding applications in fields like medicine and pollution control.
- From Board Games to Biology: The story began with AlphaGo’s success, showcasing AI’s transformative power beyond games.
- Global Impact: Over two million researchers worldwide have used AlphaFold2, democratizing access to advanced protein modelling.
I’ll be honest—I never really understood chemistry, and I’ve never been particularly good at it. But this news made me sit down, read, and actually want to understand. It’s not every day that something in chemistry catches my attention, but the 2024 Nobel Prize in Chemistry did just that.
This year, the Nobel Prize in Chemistry went to something unexpected yet amazing: neural networks! It feels like the future is truly here, with artificial intelligence leading the way.
Here’s the simple version of what the Nobel Committee said:
“Half of the prize goes to Demis Hassabis and John Jumper for using AI to solve a problem that chemists have struggled with for over 50 years: predicting the 3D structure of a protein from its amino acid sequence. Thanks to their work, we now have the structure of nearly all 200 million known proteins. The other half of the prize goes to David Baker, who developed methods to create proteins that have never existed before—many of which have brand new functions.”
And to think, this all started with a board game. Go is an ancient board game from China, played on a grid where players take turns placing stones to control territory. It’s known for its simple rules but incredibly complex strategies. Back in the 2010s, Go was thought to be one of the toughest challenges for computers. There were more possible moves in Go than atoms in the universe, and the best Go programs could only reach the fifth dan (a ranking level). That was until 2015, when Demis Hassabis and his team at DeepMind created AlphaGo—a program that used deep learning to teach itself how to play, much like how kids learn by being rewarded for good behaviour (a method called reinforcement learning).
When AlphaGo defeated Lee Sedol, one of the world’s top players, it shocked the Go community and led to South Korea investing heavily in AI development.
But Hassabis had bigger plans than just winning a board game. In 2014, Google bought DeepMind for around $500 million, and they were right to see its potential. AlphaGo’s successor, AlphaFold, developed by Hassabis and protein modelling expert John Jumper, took on a much tougher challenge: figuring out the shape of proteins. Think of proteins as tiny machines in our body, each with a unique shape that determines what it can do. AlphaFold could predict these shapes in just minutes—something that used to take scientists years, if they could even manage it.
By October 2024, over two million people across 190 countries had used AlphaFold2’s open-source code. This was a game-changer, turning years of lab work into just minutes of computer processing. It’s not just a time-saver—its helping researchers make new discoveries in medicine, environmental science, and more.
David Baker, the third Nobel laureate, took things further. He created the Rosetta neural network, which can work in reverse: starting with the desired shape of a protein and figuring out the sequence of amino acids needed to create it. In other words, it allows us to design entirely new proteins for specific tasks—like breaking down pollutants, fighting diseases, or creating new materials. Imagine designing a custom-made protein to solve a specific problem, like a key made perfectly for a lock. (Fun fact: DeepMind also launched a similar tool called AlphaProteo.)
AI keeps pushing boundaries, and today it’s not just about winning games—it’s about changing how we understand and engineer the very molecules of life. From designing new medicines to tackling climate change, the possibilities are endless. This year’s Nobel Prize proves we’re not just predicting the future—we’re building it.
image credit: ©Terezia Kovalova/The Royal Swedish Academy of Sciences
