Yes indeed. Your comment was sarcastically perfectly succinct and to the point. And it left a mark on me as being the best way to get across this "magic math" stuff encroaching into computer science. It does appear to be a very powerful tool, but with absolutely no introspection or any way to validate or invalidate content. Best one can hope, is that the dataset has no inherent biases ~ but even that's no claim the algo won't create a layer to make the biases, and then directly chose on biases.

I probably should have cited it (was on mobile, but oh well :/ ). Well, I'll do it now: https://news.ycombinator.com/item?id=14219450

If anything, please consider this copy/edit as a sign of respect, and not of mockery :)

No worries. It's all good :)

Are you confusing parameters with outputs?

Well, the numbers you get in the output are function parameters.