further more of course, I have done some interpolation on things such as panguweather when it comes to testing how accurate it is for climate prediction research though that is lesser research

Except AI is a greater catagory and it isnt a new one either. It has been called that for the last 70 years. Plus LLMs are not the only thing that uses transformer architecture either. Even something like panguweather does.

ai-models-panguweather-gfs 0.0.11 (Beta)

An ai-models plugin to run PanguWeather

Author: "European Centre for Medium-Range Weather Forecasts (ECMWF)"
🧑‍💻Repository

ai-models-panguweather 0.0.9 (Beta)

An ai-models plugin to run PanguWeather

Author: "European Centre for Medium-Range Weather Forecasts (ECMWF)"
🧑‍💻Repository

What does actual mean in this context for you because every place is not gonna implement it the same way. In fact where is the line for genAI. Like is panguweather genAI still?
That said even on a base level content summarization alone is a benefit for converting website to wga standards

Huawei’s #PanguWeather AI is now among China’s top-performing weather models—jumping from 7th in January to 2nd in recent national tests 🌦️ Powered by its Pangu-Weather system, Huawei is proving AI can rival traditional forecasting.
www.bloomberg.com/news/article...

with 1 I think you should also directly read his passage and check his links because he specifically describes that he is not saying they arent inaccurate, just that they have some basic usages such as we even discussed earlier with how panguweather works or how people use it for coding

I mean to be hoenst you cant do something like panguweather without the ability to have visual embedding. In many ways, genAI is basically just the demo of the visual embedding for things like panguweather but people like to build ont hat aspect

I mean using panguweather has been fun

advanced data-driven models, including PanguWeather and GraphCast, as well as skillful ECMWF systems. Additionally, we empirically show the effectiveness of our model designs and high-quality prediction over spatial and [6/7 of https://arxiv.org/abs/2502.19750v1]

🥉 Hakim et al., Dynamical Tests of a DL Weather Prediction Model.
This short paper evaluates whether the dynamical behavior of PanguWeather aligns with expectations, by assessing the response of the model to small perturbations of the input.

journals.ametsoc.org/view/journal...

3 AI-based models for weather forecasting ( #Panguweather, @deep-mind.bsky.social's #GraphCast
@stephanhoyer.com, @ecmwf.bsky.social's AIFS) produce skillful forecasts across different climate states, e.g. in proxy #preindustrial, present-day & future 2.9K warmer #climates.

RMSE for 2m temp: (2/7)

If I wanted to aquire some to #gpu #compute to try running either #GraphCast or #PanguWeather ML weather forecast models, how could I do that? This is a private project, so it can't cost a lot.