Thomas Kwa comments on You can, in fact, bamboozle an unaligned AI into sparing your life

[Thomas Kwa]

I appreciate the clear statement of the argument, though it is not obviously watertight to me, and wish people like Nate would engage.