As A.I. gradually increases in complexity, it becomes more weaponizable. Its ok for something like a roomba, theres no way that thing could be dangerous, but if the thing is doing the dishes for you, that thing is potentially fricken dangerous if they can repurpose the thing. Autonomous cars are in the same situation, thats why I thought openpilot (George Hotz) is such a bad idea just teaching everyone how to do it. But I guess its whats happening...