In pharmacology there is the notion of “pleiotropic” effects. It essentially means “side-effects.” However, it’s commonly used to refer to unforeseen positive side-effects. I think ML work in the open source has good pleiotropic effects for ML as a field. Not the least of which is that n00bs like myself can see it in action. It is sufficiently difficult that the only way to wrap your head around it is to see it in code and in operation… over and over again.