Bears mentioning that all so called 'conservation laws' too are 'just a summary of costly observations'.
It appears that 10% killing-eveyone-ness in a thing is just about high enough to convince everyone of its great power while simultaneously not high enough to propel anyone to take meaningful action to stop it. It's practically marketing copy for the AI labs.
And it makes a significant number of people go, "Oh! absolutely, I believe that ... but that means there's a 90% chance that it won't, and in that future, it's the person who tried to kill everyone but failed who becomes rich as Crassus. (Everyone dies equally in the 10% future.) So yeah, I'll take tho... (read more)
If everyone (for a given value of 'everyone') in China implicitly buys into the Three-body Problem's "Dark Forest" in which any and all 'others' are out to get you by definition, shouldn't that make them more, and not less amenable to agreeing with the statement that "AI will also be out to get us, just as soon we enable it to"?
What else do they believe that blocks them from this next step?