A former leader of the US National Oceanic and Atmospheric Administration has cautioned that artificial intelligence, for all its advances in weather forecasting, cannot replace the data gathered by hurricane hunter aircraft flying directly into storms — a warning with particular resonance for Bermuda as the island prepares for the Atlantic hurricane season.

The former NOAA official argued that while AI-driven models have improved dramatically in recent years, they still depend on real-world observations collected inside tropical cyclones. Hurricane hunter crews, he noted, drop instruments into storms to measure wind speed, pressure and temperature at levels that satellites and computer models cannot capture on their own. That data, gathered by aircraft operated by NOAA and the US Air Force Reserve, feeds directly into the forecasts that meteorologists and emergency planners rely upon.

For Bermuda, which sits squarely in the path of many Atlantic systems, the accuracy of those forecasts is a matter of public safety. The Bermuda Weather Service and the Emergency Measures Organisation depend heavily on US forecast data when issuing warnings, and even small improvements in storm track and intensity predictions can make a significant difference in preparation time for residents and businesses.

The remarks come amid growing debate over how much government agencies should invest in emerging AI forecasting tools versus traditional observation programmes. The former NOAA leader stressed that the two should be seen as complementary rather than competing, with AI enhancing — but never eliminating — the need for aircraft and personnel willing to fly into the eye of a storm.

Forecasters are expecting another active hurricane season, and Bermuda officials have urged residents to review their emergency plans well before the peak months of August through October.