Winner of Nobel Prize in chemistry describes how his work could transform lives

The Nobel Prize in chemistry went to three scientists for groundbreaking work using artificial intelligence to advance biomedical and protein research. AlphaFold uses databases of protein structures and sequences to predict and even design protein structures. It speeds up a months or years-long process to mere hours or minutes. Amna Nawaz discussed more with one of the winners, Demis Hassabis.

Read the Full Transcript

Amna Nawaz: This year’s Nobel Prize in chemistry went to three scientists, David Baker, John Jumper, and Demis Hassabis, for their groundbreaking work using artificial intelligence to advance biomedical and protein research.

The A.I. model they developed, called AlphaFold, uses databases with hundreds of thousands of protein structures and millions of protein sequences to predict and even design protein structures, speeding up a months- or years-long process to mere hours or even minutes.

AlphaFold was rolled out just four years ago and has since been cited in scientific studies more than 20,000 times.

Joining us now from London is Demis Hassabis, co-founder and CEO of Google DeepMind and recipient of the Nobel Prize in chemistry.

Demis, congratulations and welcome.

Demis Hassabis, Co-Founder and CEO, Google DeepMind: Thanks so much. Great to be here.

Amna Nawaz: So, first and foremost, what did you think when you heard the news?

Demis Hassabis: Well, I was totally stunned, to be honest, and it still hasn’t really sunk in even now, 24 hours later, so it just feels very surreal.

Amna Nawaz: So I’m going to try here in simplest terms, which is not simple at all, but to condense the work you and your colleagues have done.

Basically, you and your colleague John Jumper discovered new and powerful ways to not only decode, but also design proteins using artificial intelligence. I’m not even going to try to understand the details of your work, but in terms of application, how could this impact future development of things like medicine and vaccines? What does it mean?

Demis Hassabis: Well, proteins are the kind of building blocks of life, really. So everything, any — I mean, all the functions in your body are kind of supported by proteins.

And it’s really important to understand their structure, their 3-D structure, so that you can understand what the function is they have. And so that’s what AlphaFold, our program, does. It predicts that 3-D structure just from the genetic sequence. And it’s going to be really important for things like drug discovery and understanding disease.

Amna Nawaz: There’s one scientist who reacted to the news of your award by calling your work the Holy Grail in terms of what it’s been able to do. Do you agree with that characterization?

Demis Hassabis: Well, it’s very nice of them to say so.

I mean, it’s certainly a grand challenge. I mean, one — the thing — the reason that I sort of — it caught my attention is, it’s been a grand challenge of biology for the last 50 years. So people have been predicting since the ’70s that this should be possible. But, until now, no one has been able to do it to an accuracy high enough that it’s useful for biologists and medical experts.

Amna Nawaz: So, just in the way of background, you co-founded DeepMind back in 2010. But before that, I was interested to read you designed video games.

Before that, you were a chess prodigy. You were once even ranked the second highest player in the world under the age of 14. How did those endeavors, chess and gaming, feed into this work today?

Demis Hassabis: Well, there’s — actually there’s a connection all the way through my career, even though I have done different things.

So it’s because of gaming and chess specifically that I got — started to think about thinking. And I was trying to improve my own thought processes, as you do when you’re playing chess for the junior teams and the national junior teams and things like that. And I was playing it very professionally.

And part of that training, you also use chess computers, very early chess computers in the ’80s. They were actually physical blocks of plastic that you had to press the keys on. And I was sort of intrigued by the fact that someone had programmed this inanimate object to actually play chess and play chess well.

And that’s what got me into A.I. So — and then I studied neuroscience, as well as computer science, try to understand better how the — how own brains work and how intelligence is produced and the mechanisms behind it. And then, finally, that all comes together with the work we have been doing on A.I.

Amna Nawaz: As you will know, one of your fellow laureates this year is a man named Geoffrey Hinton, who’s often called the godfather of A.I.

He resigned from Google last year. And he’s really been sounding the alarm on what he says are the potential dangers here, that, as he puts it, he worries that the overall consequences of this might be systems that are more intelligent than us, that might eventually take control.

Do you share that concern?

Demis Hassabis: You know, I have known Geoff. We have been colleagues for many, many years. And he’s a fantastic scientist.

And I think that my view is sort of more moderate than that. I feel like there — of course, I have worked my whole life on A.I. because I think it’s going to be unbelievably beneficial to humanity and to society. AlphaFold is just I think the first expression of that, and I think we can go on next to try and cure many terrible diseases.

I think it can help with climate crisis, new materials, new energy sources, new mathematics. I think A.I. is going to accelerate scientific discovery, medical discovery across the board. So those are just some of the benefits I think A.I. is going to bring and why I have worked my whole life on A.I.

So it’s going to be this hugely transformative technology. But as with any new powerful technology, and perhaps A.I. will be the most powerful, it comes with risks, attendant risks as well and unknowns. And some of those are to do with controllability, understanding these systems, interpreting what they do, and how to manage the — what are the values we want these systems to have? What do we want to use them for? How do we want to deploy them?

And some of these questions are technical in nature, technical challenges, and others are more societal and need discussion with the whole of society, civil society, academia, as well as the tech companies and industrial labs and also government.

And I have been encouraging all of those debates to happen. And it’s not — it’s great that it’s starting to see those things happen. And I think, given enough time and effort, we will solve these challenges. I’m a big believer in human ingenuity. And — but we need to start discussing and researching those things now.

Amna Nawaz: That is Demis Hassabis, co-founder and CEO of Google DeepMind and recipient of this year’s Nobel Prize in chemistry.

Demis, thank you and congratulations again.

Demis Hassabis: Thank you very much.