One problem is if you can train your ML correctly and not on garbage data. To predict the world you need a giant input and a ton of data. And you also have cha

One problem is if you can train your ML correctly and not on garbage data. To predict the world you need a giant input and a ton of data. And you also have chaos, a little input divergence can cause a big output difference, I am not an ML-ANN dev but chaos feels to me impossible to approximate/interpolate with ANN, I would love to know if I am wrong and you can do any meaningful stuff on chaotic systems.