Well, the thing is, they learn what they are being fed, right? If someone decides to have some fun by asking are you going to destroy humans and have a deep convo with a bot like that, that thing is going to memorize these things and learn from those chats.. I think those AI bots that actually talk about nuclear weapons and destroying human beings may be handled by people with bad intentions? I think it's quite worrysome honestly. The video was fun though lol.. But it made me think afterwards ;)