Indeed, any flow can be modeled in the form of partial differential equation system. There are bundle of studies on applications of AI in chemistry, medical, earth sciences, neurosciences, bio informatics, physics ...are to name a few.

See also some links related to your query : Article Turbulent hydrodynamics experiments in high energy density p...

https://www.springer.com/gp/book/9780792352259

Article Reynolds averaged turbulence modelling using deep neural net...

Yes exactly, thanks for the links. So in the studies of quantum field theory (QFT), the next question is: Will AI Pattern Matching Deep Learning eventually match LHC->FCC ever-higher energy density pressure collision plasma "particle resonance" peak width hydrodynamic flows with Schwinger local field differential turbulence?

I work in the field of Data Science (A.I. and Machine Learning), AIs can do almost anything but they are also not a magic solution for all problems, they can be trained to do almost anything as long as you are able to train them for it and be able to verify that the AI can do it and if it cannot do it what to change in the training set for better success (the evaluation is also very tricky part to define), AIs training is more time consuming then the AI coding in general, I usually write the code in few hours, train the AI for few days and then it solved what I want in 1 hour, this is normally a simplified answer, some training takes months or years.

Regards, Alex

Thanks Aleksandar Ari Sokolovski for describing your experience with AI machine learning: " (the evaluation is also very tricky part to define)."

Reminds me in mathematics: First define the problem. This problem is of course in the machine intelligence analysis domain - I'm guessing you coded primarily in Python, did any of your AI training solutions involve the analysis domain?