When AG5 passes us in February 2023, the Earth’s gravity will bend its orbit a little bit, changing the path the rock takes. If it passes close to the Earth the orbit changes a lot; if it’s too far the orbit changes only a little. But if AG5 passes us at just the right distance, the orbit will change just the right amount to put it on a collision course with Earth. This region of space is called a “keyhole”, and in this case, should AG5 slip through it, it will hit us 17 years later, in 2040. That collision, though not global in scope, would be catastrophic: equal to about a 100 megaton explosion, twice that of the largest nuclear weapon ever detonated.
The problem is, we don’t know the orbit of AG5 well enough to know if it will travel through the keyhole or not. As I pointed out in the article about the asteroid 2012 DA14, it can be tricky to try to predict asteroid orbits too far into the future. The orbit of an asteroid is determined by making measurements of its position over time, and because of various effects (like blurring due to our atmosphere) it is impossible to get exactly precise positions. They can be good enough to get an accurate orbit for the next few years, but the farther into the future you look, the fuzzier that path gets.
So what do we do about it? NASA asteroid expert Don Yeomans says we should wait until 2013 and observe the asteroids trajectory. Apollo 9 astronaut and astronomer Rusty Schweikart says we should start planning now to intervene and deflect the asteroid if it becomes a threat. You can read all about the asteroid and the debate over intervention at Bad Astronomy. Link