By: Emily Ao

Early July, when Hurricane Beryl first hit the Caribbean, a top European weather agency warned that the tropical storm would hit Mexico next. This prediction, made based on global observations by planes, spacecraft, and other machinery, was quickly shattered by A.I.

While weather agencies were busy analyzing collected data, artificial intelligence software rapidly generated predictions based on previously learnt information about patterns in the planet’s atmosphere. It claimed that the storm would hit Texas, and sure enough, on July 8, when Beryl hit landfall, it struck the Lone Star State.

Contrary to what was predicted by weather forecasts, Hurricane Beryl struck Texas harshly, knocking out power for millions and killing at least 38.

Experts say that this unexpectedly accurate prediction gives us a peek of A.I’s full capabilities. GraphCast, the program used to predict Beryl’s path, was created in London under a company by Google. As stated in a New York Times article, A.I. can now do what humans take hours to do in a span of a few minutes.

According to Matthew Chantry, an A.I. specialist at the European Center for Medium-Range Weather Forecasts, A.I. has the full capability to surpass his agency when predicting storm paths. Because of its efficiency, it can also swiftly pick up changes in the atmosphere, therefore leading to more accurate results.

However, instead of fully relying on A.I, researchers say that it is best to have humans on standby. This way, researchers can take the opportunity to learn from A.I instead of solely relying on technology to get tasks done.