A new era of discovery

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 in science has been in the headlines lately, but the potential of AI to advance science has long been a motivating force for many in the field, dating back to early AI researchers, such as Alan Turing and Christopher Longuet-Higgins, and to many in recent decades including my colleagues at Google DeepMind and Google Research.

The excitement around AI and science is not because of a belief that AI is a replacement for scientists, but because many confounding problems in science benefit from the use of computational techniques — thus making AI a powerful tool to assist scientists.

We saw early signs of that assistive potential with Hodgkin and Huxley’s use of computational approaches to describe how nerve impulses travel along neurons, work that would win them the Nobel Prize in 1963.

Fast forward to my colleagues Demis Hassabis, John Jumper and the AlphaFold team whose work using AI recently won the Nobel Prize in Chemistry, solving the “protein-folding problem” posed by Nobel laureate Christian Anfinsen in the 1970s.

So how is AI helping advance science?

I’ll start with speed. In some areas of science, increasingly capable AI is making it possible for us to condense hundreds or even thousands of years of research into a few years, months, or even days.

AI is also helping expand the scope of research – enabling scientists to look at many things at once — and in new ways — rather than one by one.

AI advances — along with access to insights from using it — are enabling many more people to participate in research, so that we can further accelerate scientific discovery.

AI is enabling landmark progress in multiple scientific disciplines

Let me share briefly a few examples of how AI is enabling landmark advances, starting with AlphaFold:

With AlphaFold, over the course of a year my colleagues were able to predict the structure of nearly every protein known to science — over 200 million of them. And with Alphafold 3, they have extended beyond proteins to all of life’s bio-molectures including DNA, RNA and ligands.

To date, AlphaFold has been used by more than 2M researchers in more than 190 countries, working on problems ranging from neglected diseases to drug-resistant bacteria.

AlphaMissense, which builds on AlphaFold, enabled my colleagues to categorize almost 90% of 71M possible missense variants — single letter substitutions in DNA — as likely pathogenic or likely benign. By contrast, only 0.1% have been confirmed by human experts, albeit in more detail.

When the human genome was initially sequenced — an incredible achievement — it was based on a single genomic assembly.

Last year, my colleagues in Google Research, using AI tools and working with a consortium of academic collaborators, released the first draft reference human pangenome.

This was based on 47 genomic assemblies, thus better representing human genetic diversity.

In neuroscience, a 10-year collaboration between my colleagues in Google Research, the Max Planck Institute, and the Lichtman Lab at Harvard, recently produced a nano-scale mapping of a piece of the human brain — that is a level of detail never previously achieved.

This project revealed never-before-seen structures in the human brain that may change our understanding of how the human brain works. This will perhaps lead us to new approaches to understanding and tackling neurological diseases like Alzheimer's and others. The full mapping has been made publicly available for researchers to build on

Beyond the life sciences, we’re seeing progress in other domains.

In a landmark achievement for climate modeling, we combined machine learning with a traditional, physics-based approach to build NeuralGCM.

This allows us to simulate the atmosphere more accurately and efficiently — NeuralGCM can simulate over 70,000 days of the atmosphere in the time it would take a state-of-the-art, physics-based model to simulate only 19 days.

There are other similar breakthroughs such as the work by my colleagues at Google DeepMind on GenCast, a state-of-the-art AI model that predicts weather conditions up to 15 days in advance more accurately and much faster than the industry gold-standard weather simulation system.

Our Quantum AI team is making progress on questions that previously were the realm of science fiction, like studying the characteristics of traversable wormholes.

This opens up new possibilities for testing quantum gravity theories originally posed with the Einstein-Rosen bridge almost ninety years ago.

In fact, Quantum is an area where we’re beginning to see promising bidirectional reinforcement between AI and science.

In one direction, AI is advancing our progress in quantum computing — in the other, quantum is helping advance research in AI.

There are many other such examples that we are working on in material science, fusion, mathematics and more – all of these, in collaboration with many academic scientists.

Scientific advances enabled by AI are having real world impact

Beyond such breakthroughs, AI is also advancing science in ways that are already providing tangible benefits for real people in areas like climate and healthcare.

Let me start with an example from climate adaptation. Flood forecasting is a more frequent and urgent problem due to climate change. Now, advances in AI have enabled us to fill in large gaps in data to predict riverine flooding up to 7 days in advance with the same accuracy as nowcasts. After an initial pilot in Bangladesh, our early-warning platform — Flood Hub — now covers over 100 countries and 700 million people.

And for an example in climate mitigation, consider the following: the formation of contrails has long been a known driver of emissions in aviation — accounting for as much as 35% of aviation's global warming impact.

My colleagues in Google Research developed an AI model that predicts where contrails are likely to form, and in partnership with American Airlines, tested it on 70 flights. We measured the impact and found a 54% reduction in emissions.

Similarly, AI offers much promise for disease detection. For example, eight years ago, Google researchers found that AI could help accurately interpret retinal scans to detect diabetic retinopathy, a preventable cause of blindness that affects roughly 100 million people.

We developed a screening tool that has been used in more than 600,000 screenings worldwide. And new partnerships in Thailand and India will enable 6 million screenings over the next decade.

We have been implementing other examples including in tuberculosis, colorectal cancer, breast cancer and maternal health.

The Road Ahead

Despite the progress, this is just the beginning. There’s so much still to do.

I see three key areas to focus on to fully realize AI’s potential to help advance science and bring tangible societal benefits:

First, we need to continue to make progress on AI’s current limitations and shortcomings — and to increase AI’s capabilities to be able to assist in developing novel scientific concepts, theories, experiments and more.

Second, we need a sustained commitment to the scientific method and to responsible approaches to using AI to advance science.

We need scientists, ethicists and safety experts — like many in this room — working together to address the risks most particular to science, like viruses and bioweapons, as well as challenges like bias in data sets, privacy preservation, and environmental impacts.

Third, we need to prioritize making AI-enabled research, tools and resources more accessible to more scientists in more places — and to make sure the progress we make benefits people everywhere.

I am excited about what lies ahead in this new era of discovery.

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

And there is so much we can do to enable the amazing scientists here and elsewhere in their work — we’ll hear from some of them today.