World pop. = six billion. Divided by 10,000 = 600,000, divided by avg. life expectancy (let's say 60 years) = 10,000. Are there really 10,000 people killed in airline crashes every year?
And how do they figure the chances of being killed by an asteroid - when nobody in recorded history has ever been killed by one?
Basically, they calculate the number of asteroid impacts relative to the age of the Earth.
As for the 1 in 12,000 figure: what it doesn't say is that it's an all-or-nothing issue. No asteroid impact: no deaths. 1 large asteroid impact: over six billion (or, as the late Dr. Carl Sagan would say, "six BILLL-YUN") deaths.
It's like smoking a cigar in a fireworks shop. If nothing goes wrong, you're fine. If something does go wrong, you're hosed.