I understand it’s not needed but is it wrong? I see machine learning as a general purpose black box function approximation tool. It could be a polynomial curve

I understand it’s not needed but is it wrong? I see machine learning as a general purpose black box function approximation tool. It could be a polynomial curve or something else entirely, as long as we have enough data we can approximate it using ML. So in one way I see it as a “lazy man’s easy way out” tool.