@ketchup71 they will 100% use all the things that people input there for their next round of compiling training data—because they have no other way to refine and grow their models