Accuracy beats traditional weather forecasts! Google (GOOGL.US) releases AI weather prediction model GenCast

Google (GOOGL.US) has unveiled an artificial intelligence weather forecasting model called GenCast that outperforms traditional weather forecasting models in accuracy. According to the company, GenCast forecasts weather 15 days in advance faster and more accurately than other weather forecasting systems, including the European Centre for Medium-Range Weather Forecasts (ECMWF) system. In a study published in the journal Nature, GenCast was more accurate than the ECMWF model in 97.2% of cases if the forecast range was greater than 36 hours, with an accuracy rate of 99.8%. A blog post from Google's artificial intelligence research arm DeepMind said: "Ensemble prediction expresses uncertainty by making multiple predictions of different possible scenarios. If most predictions show a hurricane hitting the same area, the uncertainty is low. But if they predict different locations, the uncertainty is higher. GenCast strikes the right balance, avoiding overconfidence or underconfidence in its predictions." GenCast's forecasting speed is also much faster than traditional weather models. For example, a single Google Cloud TPU v5 can generate 15-day forecasts in 8 minutes, while producing more forecasts. In contrast, traditional physics-based ensemble predictions take hours on supercomputers. Google said the improved forecasting would benefit agriculture and help predict extreme weather events such as typhoons and hurricanes. "We are eager to work with the broader weather community, including academic researchers, meteorologists, data scientists, renewable energy companies, and organizations focused on food security and disaster response," Google said.