The team found that GenCast outperformed the company’s previous AI weather program, which launched in late 2023 and provided a reliable 10-day forecast.
Google DeepMind has unveiled a new artificial intelligence tool, claiming it has unmatched capabilities in generating 15-day weather forecasts with exceptional speed and accuracy. In a recent report published in Nature Journal on Wednesday, the London-based company asserted that its new AI model could surpass the best global forecasts used to track dangerous storms and save lives.
Ilan Price, the lead author of the study and a senior research scientist at DeepMind, highlighted that the new AI agent, ‘GenCast’, is significantly faster than traditional forecasting methods. “And it’s more accurate,” he added.
The team found that GenCast far outperformed the company’s previous AI weather program, which was launched in late 2023 and provided a reliable 10-day forecast.
“I’m a little reluctant to say it, but it’s like we’ve made decades of progress in just one year. We’re seeing incredibly fast advancements,” remarked Rémi Lam, the lead scientist on the earlier project and one of the co-authors of the new study.
The European Center for Medium-Range Weather Forecasts is regarded as a global leader in atmospheric prediction, consistently delivering the most accurate projections in comparative tests.
DeepMind’s new AI program was tested against the European Centre for Medium-Range Weather Forecasts’ Ensemble Prediction System, which is used by 35 countries worldwide to produce weather forecasts.
In the study, the team compared how the 15-day predictions from both systems performed when forecasting a specific set of 1,320 global wind speeds, temperatures, and other atmospheric variables.
The results showed that DeepMind’s new AI agent outperformed the centre’s forecasts 97.2% of the time. The authors of the study argue that this accomplishment “marks the beginning of a new era in operational weather forecasting.”
“This is a significant achievement… an important step forward,” Kerry Emanuel, a professor emeritus of atmospheric science at the Massachusetts Institute of Technology, told The New York Times. In 2019, Dr. Emanuel and six other experts had highlighted the “enormous socioeconomic benefits” of extending reliable weather forecasts from 10 days to 15 days, as it could help mitigate the impact of extreme weather events.
