Why AI has trouble predicting the fury of hurricane intensity

A researcher explains why AI weather models, despite rivaling top physics-based systems globally, struggle specifically with rapid hurricane intensification, citing Hurricane Polo's jump from tropical storm to Category 5 with 180 mph winds in 24 hours off Mexico's Pacific coast in September 2026. The gap traces to data: global datasets are rich, but the open-ocean, three-dimensional detail needed to train models on fast-developing storms is sparse, pieced together from uneven satellite coverage and imperfect simulations.

The piece also raises a deeper limit: hurricanes may contain genuine chaotic behavior, meaning tiny differences in starting conditions can snowball unpredictably, which would cap how accurate any model, AI or otherwise, can ever get at long forecast ranges.

The full dispatch is available from the source below.