1) Something is going catastrophically wrong with the plane. It's flying now, but soon that fire will burn through the hydraulics, safer to punch while it's still in controlled flight.
2) There is not enough fuel to put it on a runway. Once again, punch while it's still stable. We've even seen a civilian do that--ferry flight with one of those planes with an integrated parachute. He had a fancy rig with extra fuel in the cabin, it wouldn't feed. Without it he wasn't going to make Hawaii.
3) We have seen a Navy pilot correctly punch from a fully functional aircraft. He was on final when somebody launched an SM-2 at him. Low, slow, defenses off--no chance, he punched. He was pointed right at the carrier at the time, having an autopilot do something else would be a good idea. (There would always be a chance that the missile was destructed in time.)
And with semi-functional planes:
4) We have seen an Israeli pilot bring their bird back with one wing. He had to land very hot but there was a long enough runway, he was able to do it. But what if it's even worse? There can be enough damage that your minimum airspeed is above any runway you can reach or above what your wheels can handle. Or maybe the Navy would prefer the pilot to eject rather than risking a major mishap on the deck.
5) What if the problem is elsewhere? We have seen a pilot punch from an apparently-functional F-35. The problem was actually in his controls. (Yes, he has been found wrong--the first two investigations cleared him (if the plane is not responding properly to pilot inputs below 6,000' AGL, eject), they finally found a panel that would declare him wrong. Doesn't make him wrong.)
I also recall reading about another damaged A-10 in the same conflict that managed to touch down only to discover the brakes didn't work. Pilot overran the runway but survived. Can't recall pilot's name off the top of my head.
Compare to 20 in 20 jet airplane crashes resulting in death and suddenly pulling that lever might seem a worthwhile risk to take
But more than that result in injury. The possible injuries are pretty severe.
> Compare to 20 in 20 jet airplane crashes resulting in death
Crashes if the plane is totally uncontrollable, probably yes.
But there's a lot of gray area in between "totally uncontrollable" and "controllable enough that an autopilot can fly the plane". There are plenty of cases where a pilot was able to make a controlled enough crash that they walked away from it, even though the plane itself was totalled.
And once we get to the point of "controllable enough that an autopilot can fly the plane", the pilot would have no reason to eject--because the plane is controllable enough that the autopilot can fly it. Which means whatever problems exist can't be very severe--or the autopilot would be disengaging, because it needs things to be working pretty well to fly the plane at all. That was the point of my response in the GP to this post.
E.g, if it's already been programmed to fly straight and level, continue to do that. If it's deactivated, stay deactivated.
Just seems like a whole 'nother set of characteristics to test otherwise, as well as adding extra unpredictability. The aircraft is probably damaged / on fire, so its flight characteristics are already going to be extremely different to normal. The best thing in the moment may be to let the aircraft lawn-dart in a field, rather than attempt to get straight and level, and in the process potentially fly over inhabited area or towards a friendly set of aircraft / buildings / vehicles.