Indonesian Political, Business & Finance News

Google Warns of Faster Disaster Predictions After 700,000 Deaths

| Source: CNBC Indonesia Translated from Indonesian | Technology
Google Warns of Faster Disaster Predictions After 700,000 Deaths
Image: CNBC Indonesia

Jakarta, CNBC Indonesia - Predicting the emergence of deadly tropical storms or cyclones no longer takes a long time. Google DeepMind and Google Research have officially launched their latest artificial intelligence (AI) technology called WeatherNext, which is claimed to be able to predict the direction and strength of storms one day earlier than current meteorological technology.

A research report published in the journal Nature states that the leap in accuracy produced by WeatherNext is equivalent to a decade of progress in meteorological technology.

“On average, our model provides forecasters with an extra day of prediction time. Our three-day forecast is equivalent to what previous models could only provide for two days ahead,” wrote the Google research team, quoted from the official Google DeepMind blog, Friday (7/8/2026).

This step is crucial given that tropical cyclones such as hurricanes and typhoons are among the deadliest natural disasters. Over the past 50 years, these phenomena have claimed more than 700,000 lives and triggered economic losses of up to US$1.4 trillion (approximately Rp25,000 trillion) globally.

The technology was developed through massive collaboration between Google AI researchers and leading weather institutions worldwide, including the US National Hurricane Center (NHC), the UK Met Office, and the Cooperative Institute for Research in the Atmosphere (CIRA).

WeatherNext’s capabilities are not merely theoretical. Throughout the 2025 storm season, the AI model was used by the NHC to predict Hurricane Melissa in Jamaica with precision, including the potential for rapid intensification and its landfall point.

The early information provided vital time for rescue teams on the ground to carry out evacuations and emergency preparations before the storm struck.

Technically, WeatherNext is able to solve a classic problem in weather science. Until now, predicting storm tracks required global-scale atmospheric modelling, while predicting storm strength required very high-resolution local modelling. WeatherNext has succeeded in combining both capabilities in a single AI model.

Remarkably, the AI is able to run 1,000 weather simulation scenarios in less than one minute using Google TPU chips.

In an effort to reduce the impact of the climate crisis and natural disasters, Google has decided to make the source code for WeatherNext 2 and WeatherNext Cyclones freely available as open source.

The move is expected to help meteorological agencies, researchers, and non-profit organisations in various countries, including developing nations, to strengthen early warning systems for disasters, support the renewable energy sector, and anticipate extreme weather.

The general public and researchers can now also monitor storm movements and global weather predictions directly through the Weather Lab visualisation platform integrated into the Google Earth AI ecosystem.

More from 7 August 2026 →

Cite this article
Source: jawawa.id (https://jawawa.id/newsitem/google-warns-of-faster-disaster-predictions-after-700-000-deaths-1786096370)
Originally published by CNBC Indonesia (https://www.cnbcindonesia.com/tech/20260807121851-37-757398/700000-orang-tewas-google-peringatkan-bencana-besar-lebih-cepat), translated from Indonesian
Print