1. Meeting the “grand challenge” of 50 years of protein structure prediction
Experts have described the demystification of protein folding as a “grand challenge” for decades. In 2022, Google DeepMind shared the predicted structures of 200 million proteins from their AlphaFold 2 model. Previously, it typically took a year or more to determine the 3D structure of a single protein. AlphaFold can predict these shapes with remarkable accuracy in minutes. By publishing protein structure predictions in a free databasethis has enabled scientists around the world to accelerate progress in areas such as development new drugs, fight against antibiotic resistance And fight against plastic pollution. In the next step, the AlphaFold 3 model builds on AlphaFold 2 to predict the structure and interaction of all life’s molecules.
2. Show the human brain in unprecedented detail, to support health research
Few things have remained more mysterious over time than the human brain. Developed over 10 years of connectivity As part of this research, Google teamed up with others, including Harvard’s Lichtman Lab, to map a tiny part of the human brain at a level of detail never before achieved. This project, released in 2024, revealed new structures within the human brain. And the full dataset, including AI-generated annotations for each cell, has been made public to help speed up research.
3. Save lives with accurate flood forecasting
When Google’s flood forecasting project began in 2018, many believed it was impossible to accurately provide large-scale flood forecasts, given the scarcity of data. But researchers were able to develop an AI model that achieved reliability in forecasting extreme river events in ungauged watersheds within up to five days with reliability equal to or better than nowcasts ( time limit of zero days). In 2024, Google Research expanded this coverage to 100 countries and 700 million people worldwide – and improved the AI model so that it provides the same accuracy within seven days as the previous model had five days .
4. Detect wildfires earlier to help firefighters stop them faster
Wildfires are increasingly disrupting communities around the world due to hotter, drier climates. In 2024, Google Research partnered with the U.S. Forest Service to develop FireSat, an AI model and new global constellation of satellites specifically designed to detect and track room-sized wildfires. class by delivering higher resolution images in 20 minutes. This will allow fire services to respond more quickly, potentially saving lives, property and natural resources.
5. Predict the weather faster and more accurately
In 2023, Google DeepMind launched and open sourced the model code of GraphCasta machine learning research model that predicts weather conditions up to 10 days in advance more accurately and much faster than the industry’s benchmark weather simulation system (HRES). GraphCast can also predict cyclone trajectories (and associated risks like flooding) with greater accuracy, and accurately predicted Hurricane Lee would arrive in Nova Scotia three days before traditional models.
6. Advancing the frontiers of mathematical reasoning
AI has historically struggled with complex math due to a lack of data and reasoning skills. Then, in 2024, Google DeepMind announced AlphaGeometryan AI system that solved complex geometry problems at a level close to a human Olympiad gold medalist – a breakthrough in AI performance and the search for more advanced general AI systems. The later model formed by Gemini, AlphaGeometry 2, was later combined with a new AlphaProof modeland together they have solved 83% of all historic International Mathematical Olympiad (IMO) geometry problems over the past 25 years. By demonstrating AI’s growing ability to reason and potentially solve problems beyond current human capabilities, it has brought us closer to systems capable of discovering and verifying new knowledge.
7. Use quantum computing to accurately predict chemical reactivity and kinetics
Google researchers worked with UC Berkeley and Columbia University to perform the largest chemical simulations ever on a quantum computer. The results, published in 2022were not only competitive with classical methods, but they also did not require the heavy error mitigation typically associated with quantum computing. The ability to perform these simulations will provide even more accurate predictions of chemical reactivity and kinetics, providing a precursor for applying chemistry in new ways to help solve real-world challenges.
8. Accelerate materials science and the potential for more sustainable solar cells, batteries and superconductors
In 2023, Google DeepMind announced Graph Networks for Materials Exploration (Gnome)a new AI tool that has already discovered 380,000 materials stable at low temperatures, according to simulations. As our world seeks new approaches to energy, processing power and materials science, this work could pave the way to better solar cells, batteries and potential superconductors. Additionally, to make this technology benefit everyone, Google DeepMind has made GNoME’s most stable predictions available through the Materials Project on its open database.
9. Take a significant step toward nuclear fusion – and abundant clean energy
As the old joke goes, “Fusion is the energy of the future – and it always will be.” Controlling and using the energy that powers the stars (including our own sun) is beyond the realm of science. In 2022, Google DeepMind announced that she has developed an AI capable control the plasma inside a nuclear fusion reactor autonomously. By collaborating with EPFL’s Swiss Plasma Center, Google DeepMind built the first Reinforcement Learned system capable of autonomously stabilizing and shaping plasma in an operational fusion reactor, opening new avenues toward stable fusion and abundant clean energy for all.