most of the current AI "breakthroughs" are just people rehashing old research and pretending to have invented something new. wake me up when someone actually figures out how to make a neural network that generalizes to a real-world problem without