How to Train a Neural Net

i often think that the way we're doing neronet training these days, with the amount of data points, the amount of processing timet they require is more like re running. Like training a neural net an also be modelled as search. So we're kind of just deferring the problem, right? Like the learning part is now this brute force basically search. Ah, and we just have the resources to do it. It scares me.