Medicine was VERY unscientific until recent days. Bacteriological theory was unscientific until it wasn't. Psychiatry was, and I'd argue, still unscientific. And machine learning is equal parts of magic and statistical math. Most practitioners in ML/AI area just tweak knobs and hope that some hyperparameter does infinitesimally better.
The key takeaway is that not even knowing the appropriate metrology to engage in makes this a very hard scientific endeavour. But that is the first problem to solve - how do you measure whatever that is. And once you can measure, then you can manipulate.
There's too many of these stories Ive heard from family, friends, myself, and others to think that this area is some sort of a delusion and fake. But what is its nature? Unsure. That's why I explore.
