“But—” said Professor McGonagall. She glanced at Mad-Eye Moody, who was nodding his approval, and at Severus, who was keeping his face decidedly blank. “Mr. Potter, you just stunned Mad-Eye Moody! The most famous Dark wizard hunter in the history of the Auror Office! That should’ve been impossible!”
Moody let out a dark chuckle. “What’s your answer to that one, kid? I’m curious.”
“Well...” Harry said. “First of all, Professor McGonagall, neither of us were fighting seriously.”
“Neither of you?”
“Of course,” Harry said. “In a serious fight, Mr. Moody would’ve dropped all my copies immediately without waiting for them to attack. And on my side, if it was actually necessary to take down the most famous Auror in the history of the office, I’d get Headmaster Dumbledore to do it for me.
and then
“You might say that Mr. Moody was testing me to see if I would try to fight him, or try to win. That is, whether I’d carry out the role of somebody fighting—use standard spells I already knew, even though I didn’t expect the consequences of that action to be victory—or if I’d search through unusual plans until I found something that could win.
Yes, this idea pops up every now and then. Does anyone (perhaps UCLA alum) know of a non-offputting way to get in touch with him? I think this community has some good communicators who could explain some of the more mathematically interesting parts of the alignment problem to him.
and then
I wonder if there is some wisdom here related to AI alignment. Maybe the goal should be to get Terry Tao to do it.
Yes, this idea pops up every now and then. Does anyone (perhaps UCLA alum) know of a non-offputting way to get in touch with him? I think this community has some good communicators who could explain some of the more mathematically interesting parts of the alignment problem to him.