It is quite astonishing, if that is true. AI is used especially to identify these sorts of things. It seems the perfect fit for AI.
No you missed my point.
The meteorologists actually do know better (as opposed to what you may think).
Use of neural networks (or AI as you refer to it) in climate or even weather forecasting is something being chased by what I think is the IT sector.
Folk falling into the big data camp.
There are many uses for such techniques, but describing natural systems is not one of them.
Particularly the climate system, which includes oceans, land and atmosphere.
These big data people want desperately to get into the atmospheric sciences because they feel they can solve everything using a technique that they may think is new. It's really not, but advances in computing power have brought it to the forefront.
They think, as you, that it's a perfect fit. It's really not. And it takes precious time convincing someone from outside the field that it can't work the way they think it can. All you are doing is massaging the data to squeeze out a prediction without a physical basis.
This also brings me to a point that the models are much more useful than for just predicting something. They are used for actually figuring why a thing happened.