When Helene made landfall in Florida earlier this year, 234 people lost their lives to the worst hurricane to strike the US mainland since Katarina in 2005. It’s natural disasters like that, and their growing intensity due to climate change, that have pushed scientists to develop more accurate weather forecasting systems. On Wednesday, Google’s DeepMind division announced what may go down as the most significant advancement in the field in nearly eight decades of work.
In a post on the Google Keyword blog, DeepMind’s Ilan Price and Matthew Wilson detailed GenCast, the company’s latest AI agent. According to DeepMind, GenCast is not only better at providing daily and extreme weather forecasts than its previous AI weather program, but it also outperforms the best forecasting system in use right now, one that’s maintained by the European Center for Medium-Range Weather Forecasts (ECMWF). In tests comparing the 15-day forecasts the two systems generated for weather in 2019, GenCast was, on average, more accurate than ECMWF’s ENS system 97.2 percent of the time. With lead times greater than 36 hours, DeepMind’s was an even better 99.8 percent more accurate.
“I’m a little bit reluctant to say it, but it’s like we’ve made decades worth of improvements in one year,” Rémi Lam, the lead scientist on DeepMind’s previous AI weather program, told The New York Times. “We’re seeing really, really rapid progress.”
GenCast is a diffusion model, which is the same tech that powers Google’s generative AI tools. DeepMind trained the software on nearly 40 years of high-quality weather data curated by the European Center for Medium-Range Weather Forecasts. The predictions the new model generates are probabilistic, meaning they account for a range of possibilities that are then expressed as percentages. Probabilistic models are considered more nuanced and useful than their deterministic counterparts, which only offer a best guess of what the weather might be like on a given day. The former also harder to create and calculate.
Indeed, what’s perhaps most striking about GenCast is that it requires significantly less computing power than traditional physics-based ensemble forecasts like ENS. According to Google, a single one of its TPU v5 tensor processing units can produce a 15-day GenCast forecast in eight minutes. By contrast, it can take a supercomputer with tens of thousands of processors hours to produce a physics-based forecast.
Of course, GenCast isn’t perfect. One area the software could provide better predictions on is hurricane intensity, though the DeepMind team told The Times it was confident it could find solutions for the agent’s current shortcomings. In the meantime, Google is making GenCast an open model, with example code for the tool available on GitHub. GenCast predictions will also soon make their way to Google Earth.