"New Discovery Era Begins"

Editor’s note: Today in London, Google DeepMind and the Royal Society co-hosted the inaugural AI for Science Forum, which brought together Nobel laureates, the scientific community, policymakers, and industry leaders to explore the transformative potential of AI to drive scientific breakthroughs, address the world's most pressing challenges, and lead to a new era of discovery.

Google’s Senior Vice President for Research, Technology and Society, James Manyika, delivered the opening address; what follows is a transcript of his remarks, as prepared for delivery.

AI's impact on science has been making waves lately, but the idea of using AI to push scientific boundaries isn't new. It goes back to pioneers like Alan Turing and Christopher Longuet-Higgins, and more recently, my colleagues at Google DeepMind and Google Research have been at the forefront.

The buzz around AI and science isn't about replacing scientists; it's about how AI can tackle those tricky problems that benefit from computational power. Think of AI as a trusty sidekick for scientists.

We got a glimpse of this potential way back when Hodgkin and Huxley used computational methods to explain how nerve impulses travel along neurons. Their work snagged them the Nobel Prize in 1963.

Fast forward to today, and my colleagues Demis Hassabis, John Jumper, and the AlphaFold team used AI to crack the "protein-folding problem" that Christian Anfinsen posed in the 1970s. Their efforts earned them the Nobel Prize in Chemistry.

So, how exactly is AI helping to advance science?

Let's talk about speed first. In some fields, AI is speeding up research that would normally take centuries, squeezing it into just a few years, months, or even days.

AI is also broadening the scope of research, allowing scientists to explore multiple things simultaneously and in fresh ways, rather than one at a time.

Thanks to AI, more people can now join the research party, which means we can ramp up scientific discovery even faster.

AI is driving major progress across various scientific fields

Let me give you a few examples of how AI is making big waves, starting with AlphaFold:

In just a year, my colleagues predicted the structure of nearly every known protein — over 200 million of them. And with AlphaFold 3, they've gone beyond proteins to include all of life's biomolecules like DNA, RNA, and ligands.

AlphaFold has been a game-changer for over 2 million researchers in 190 countries, tackling everything from neglected diseases to drug-resistant bacteria.

AlphaMissense, built on AlphaFold, helped categorize nearly 90% of 71 million possible missense variants — those single-letter DNA changes — as likely harmful or benign. That's a huge leap, considering only 0.1% had been confirmed by human experts.

When the human genome was first sequenced, it was based on a single assembly. But last year, my colleagues at Google Research, along with academic collaborators, released the first draft of a reference human pangenome. This was based on 47 assemblies, giving us a better picture of human genetic diversity.

In neuroscience, a decade-long collaboration between Google Research, the Max Planck Institute, and Harvard's Lichtman Lab produced a nano-scale map of a piece of the human brain. This level of detail was unprecedented and revealed new structures that could change how we understand the brain. This could lead to new ways of tackling neurological diseases like Alzheimer's. The full map is now available for other researchers to explore.

Beyond life sciences, we're seeing breakthroughs in other areas too.

In climate modeling, we combined machine learning with traditional physics-based methods to create NeuralGCM. This model can simulate over 70,000 days of the atmosphere in the time it would take a traditional model to simulate just 19 days.

Another example is GenCast, developed by my colleagues at Google DeepMind. It's a top-notch AI model that predicts weather up to 15 days ahead more accurately and faster than the industry standard.

Our Quantum AI team is exploring what used to be sci-fi territory, like studying traversable wormholes. This could help test quantum gravity theories first proposed with the Einstein-Rosen bridge nearly ninety years ago.

In fact, quantum and AI are starting to help each other out. AI is advancing quantum computing, and quantum is helping push AI research forward.

We're also making strides in material science, fusion, mathematics, and more, all in collaboration with academic scientists.

AI-driven scientific advances are making a real-world impact

Beyond these breakthroughs, AI is also improving science in ways that directly benefit people, especially in areas like climate and healthcare.

Take climate adaptation, for example. Flood forecasting is becoming more critical due to climate change. Thanks to AI, we can now predict riverine flooding up to 7 days in advance with the same accuracy as current predictions. Our early-warning platform, Flood Hub, started in Bangladesh and now covers over 100 countries and 700 million people.

For climate mitigation, consider contrails, which contribute up to 35% of aviation's global warming impact. My colleagues at Google Research developed an AI model to predict where contrails might form. After testing it on 70 flights with American Airlines, we saw a 54% reduction in emissions.

AI also shows promise in disease detection. Eight years ago, Google researchers found that AI could help interpret retinal scans to detect diabetic retinopathy, a preventable cause of blindness affecting about 100 million people. We developed a screening tool that's been used in over 600,000 screenings worldwide. New partnerships in Thailand and India will enable 6 million screenings over the next decade.

We're also working on other areas like tuberculosis, colorectal cancer, breast cancer, and maternal health.

The Road Ahead

Despite all this progress, we're just getting started. There's still a lot to do.

I see three key areas to focus on to fully harness AI's potential for advancing science and bringing real benefits to society:

First, we need to keep working on AI's limitations and increase its ability to help develop new scientific concepts, theories, and experiments.

Second, we must stay committed to the scientific method and use AI responsibly. Scientists, ethicists, and safety experts need to work together to tackle risks specific to science, like viruses and bioweapons, as well as challenges like data bias, privacy, and environmental impacts.

Third, we need to make AI-enabled research tools and resources more accessible to scientists everywhere, ensuring that the progress we make benefits people around the world.

I'm excited about what the future holds in this new era of discovery.

There's so much we can do together to build tools that help advance science for everyone's benefit.

And there's so much we can do to support the amazing scientists here and around the world — we'll hear from some of them today.