As soon as Hurricane Isaias formed, meteorologists at the National Hurricane Center took a big gamble. They shaped their forecasts to closely match an AI hurricane model developed by Google DeepMind, rather than relying on the classic physics-based models that many Americans are more familiar with, such as European and GFS. So far, the gamble has paid off.
Forecast for this storm reflects improving reliability AI weather models as a whole, and signal a surprisingly rapid shift to the routine use of such products by human forecasters.
Relying heavily on DeepMind meant NHC forecasts Isaias would become a Category 2 on Tuesday’s first notice, when it didn’t even have a name yet. At that point, many physics-based computer models, which rely on mathematical equations describing how the atmosphere works to simulate future weather conditions, vacillated between different projections, but generally predicted a weaker storm that would take a different trajectory than that shown by DeepMind and other AI models.
If the forecast proves correct, it will mark a huge victory for Google and help those in the path of the storm, who will have had more time to prepare for a much stronger blow. Hurricane Isaias is the first hurricane of the Atlantic tropical season, largely due to the effects of a super El Niño in the Pacific, setting a record for the last first hurricane of the season in the ocean basin.
NHC forecasters’ confidence in the Google DeepMind model in particular – the latest iteration of which is WeatherNext 3 – stems largely from experience. Last season, the Hurricane Center correctly anticipated the rapid intensification of Hurricane Melissa nearly three days before its devastating landfall in Jamaica, based largely on DeepMind projections. This also comes because the Hurricane Center partnered with Google to help train the model and learn more about its capabilities.
As is currently the case in many fields, the world of meteorology is experiencing rapid progress thanks to AI, although these models are used in conjunction with traditional physical models, which still have their advantages. AI models are trained on past weather events and, in the case of DeepMind, also receive current weather data.
⛅ Get CNN Weather delivered to your inbox
- Forecasts are just the beginning. We’ll send you expert reports and stories about the weather so you always know more than just the number. Subscribe to the newsletter
DeepMind performed extremely well last season, according to NHC Director Michael Brennan, and accurately anticipated the intensity of many storms in the Pacific Ocean this season.
“The AI models showed more consistency” from one model run to the next, Brennan said of using DeepMind for Hurricane Isaias.
“It’s probably a combination of what we’ve seen in looking at these physics-based models and some inconsistencies and then just the performance of the AI guidance so far, and the fact that it’s proven to be quite reliable,” Brennan said.
Part of that evidence comes from a courageous call Hurricane Center forecasters made last year during Hurricane Melissa.
Last season, for the first time, Hurricane Center forecasters explicitly predicted that Hurricane Melissa would reach Category 5 intensity. when it was only an 80 mph Category 1 hurricane, largely based on DeepMind’s clear projections. In fact, by some measures, DeepMind even outperformed human forecasters during the 2025 Atlantic hurricane season.
With global climate change allowing for rapid intensification of tropical cyclones, the Hurricane Center has made more optimistic early forecasts for some storms, including Isaias. Before the AI entered the forecast phase, the NHC had used a rather step-by-step approach, gradually increasing a storm’s intensity ceiling as it accelerated, often putting official forecasts behind schedule.
This practice appears to have ended. Rapidly intensifying storms are now more common and the latest models are better able to anticipate them.
“We’ve definitely been doing more aggressive intensity forecasts overall over the last few years, and we definitely have more confidence in trying to do some of these more aggressive intensity forecasts early on, like we did for Melissa last year, but we also did for Helene and Milton,” Brennan said.
He cited the support of physical computer models for rapid intensification as well as weather forecasters’ understanding of the environment in which a storm forms as additional factors in quickly predicting rapid intensification.
“It’s not just one thing, but AI models have certainly performed well in some of these cases,” Brennan said. “So this is another tool in the toolbox, so to speak, or another piece of evidence that can help support these more aggressive intensity predictions,” he said of the AI models.
Brennan said forecasters don’t hesitate to rely on DeepMind, in part because they are familiar with the training data that has been associated with it.
“I think the most important thing is to make sure the models are trained on the most accurate data we have,” he said, “and we’ve worked with Google to help them train on our best track analysis, and that’s one of the reasons why I think their performance has improved in terms of intensity prediction compared to some of the previous AI models that worked well for track, but not for intensity.” These models were trained on large-scale data sets that were not specific to hurricanes, he said.
Meanwhile, Hurricane Isaias is heading toward the northeastern Gulf Coast and is expected to make landfall as a hurricane, according to the consensus of most computer models — not just DeepMind.
Gn headline
