He predicts the weather with machine learning - Dmitrii Kochkov - Young American Scientists 2026

[00:00:00]I think one of the challenges in tackling uh a lot of the extreme weather events is that making reliable weather forecasts especially at longer time horizons is extremely challenging and I think with advancements uh in you know AI based modeling of such systems the whole process will become a lot easier and allow people to take the expertise that they have and apply them to real world problems and have a you know enormous impact on uh on human lives.

[00:00:34]My name is Dimma. Uh I work at Google research uh specifically on AI for science or AI for weather forecasting and uh climate modeling. So the project that I'm working on at the moment is to advance how well we can do modeling of weather and uh climate systems, earth system in general. We know that the climate is warming and uh when we talk about uh extreme events or various you know likelihood of catastrophic events we actually don't know what that likelihood is like what's the probability that a cyclone is going to hit certain make make landfall somewhere uh or what's a once in aundred year storm event um and uh I do think that the type of models that we're building might be able to uh provide provide some insight into uh some of these questions.

[00:01:25]Extreme weather events actually are quite peculiar because uh even though they're rare, they uh affect humanity in a most severe way often times. And uh knowing those outliers or being able to predict them or even say statistically how likely uh certain outcomes are is extremely important for preparedness and risk management. The thing that excites me the most about this work is that it has such clear tangible applications where as you uh figure out how to model these systems better, uh you can enable these downstream applications that help people you know in farming or uh catastrophe planning or other industries and uh can benefit the entire the entire world.