Demis Hassabis is one of the researchers at Google's Deepmind division. There, he works on building systems that can understand data and find structures in it using AI algorithms.
He has tried to make a 50-year-old theory a reality, namely that it should be possible to predict 3D structures of proteins – one of life's building blocks – based on amino acid sequences.
It has been a challenging computational biology to do so. We used all modern machine learning techniques and the known structures that have been painstakingly discovered over the last 50 years, and we managed to build a system that with high accuracy could fold and find the structure of almost all proteins that science knows.
"Very surrealistic"
When TT reaches him a few minutes after the announcement on Wednesday morning, he has not yet had time to digest the news.
It's very difficult to take in, to be honest. It's very surrealistic, it's hard to grasp.
Although it has been speculated that the so-called Alphafold project might be awarded this year, he did not quite believe it.
You never know. It's not something you think about every day. And it wasn't so long ago that we got this breakthrough. But apparently, many have made use of Alphafold's predictions. It's clearly incredible to see what significance they have gained.
Cited thousands of times
So far, it's about basic research that has not yet had a major practical application. But the study, which was published in Nature 2021, has been cited 27,000 times, which says something about the impact.
I hope that in the coming years, we will see some real applications, such as new medicines being discovered, partly thanks to Alphafold and the structures it predicts, says Hassabis.
Ahead lies Stockholm and a place of honor at the Nobel banquet later this year.
I've seen it on TV before, and it seems to be a fantastic event. So I'm really looking forward to it.
Demis Hassabis was born in 1976 in London and is one of the founders and CEOs of DeepMind, an AI company that Google acquired in 2014. Deepmind has published over 1,000 scientific studies.
Besides cracking protein folding, Deepmind gained attention by creating the first program that beat the world champion in the board game Go.
Hassabis was named one of the most influential people in AI by Times this year, and has also written the code for the classic theme park game "Theme park", and founded the game studio Elixir Studios.
He was also knighted this year, and can therefore title himself "Sir".
Source: Times, Linkedin, The Royal Society
