This has been in the news recently. It's a good read for anyone interested in weather forecasting & developments in machine learning.

Non-expert articles:
https://deepmind.google/discover/blog/graphcast-ai-model-for-faster-and-more-accurate-global-weather-forecasting/
https://www.wired.com/story/google-deepmind-ai-weather-forecast/
https://www.nature.com/articles/d41586-023-03552-y

Published paper in Science (for the technical read):
https://www.science.org/doi/10.1126/science.adi2336

---

---
CPDN Visiting Scientist

It's important to point out that in order to train the Google model, they used atmospheric data 'reanalyses' from ECMWF's data assimilation system, which ingests huge amounts of measured data. The ECMWF assimilation system is highly complex and makes use of their IFS forecast model and requires significant compute power & storage.

Very interesting. I had heard some of the reports on the news about this but reading in a bit more detail makes me wonder if in a few years down the line, CPDN will be using AI for its work and if so will there still be a place for it using BOINC?

Hard to say. There's been discussion about the potential for machine learning applications in CPDN for some time, but it's finding the right problem to solve and the right applications.

Very interesting. I had heard some of the reports on the news about this but reading in a bit more detail makes me wonder if in a few years down the line, CPDN will be using AI for its work and if so will there still be a place for it using BOINC?

Dave, you might also be interested in this recent announcement about a new AI & environment training centre at U.Oxford:
https://eng.ox.ac.uk/news/new-training-centre-will-bridge-gap-between-environmental-science-and-ai/