F-35 can’t be found after pilot ejected

Stealth technology shrinks radar signature, meaning all else being equal, a radar pointed at a stealth aircraft will return a substantially weaker signal than for a non-stealth aircraft, particularly at certain wavelengths it's optimized to be stealthy in. Thus a system of radars that might have no gaps a non-stealth aircraft could fly through might have gaps where it couldn't detect a stealth aircraft, or radars with the resolution to accurately locate the aircraft might not be able to distinguish it from background noise. Generally very large, long wavelength radars can detect that a stealth aircraft is present, but can't accurately determine location. The only time an air defense system has ever shot down a stealth aircraft was during the 90s when a Yugoslavian force shot down an F-117, but they relied on the fact that the plane was flying along a known flight route. Under ideal circumstances defense against stealth is theoretically possible, but a system that can reliably defend against stealth under arbitrary conditions is at best impractical.

Now that's the case for a radar deliberately set up to try and detect military aircraft flying through some area. The US isn't particularly expecting any foreign nation to fly a small sortie over North Carolina any time soon. Radar coverage in the US is limited; airports have radars that can detect very non-stealthy civilian aircraft a few dozen miles away, and there are many radars at military bases and weather stations and such scattered around, but air traffic monitoring relies heavily on transponder signals for most of the air space. While a foreign adversary probably shouldn't count on the ability to overfly the US completely undetected, we almost certainly would not be able to track it along its flight path.

Of course the US does not rely on detecting foreign aircraft to protect against an attack on the mainland, but instead practices a strategy of deterrence.

But modern radar is much more sensitive, and the noise-handling algorithms better - basically it's much more able to distinguish "I hit some trees with my beam" versus "metal". Combine that with modern transmitters which can also produce tighter beams and all the other electronic goodness, and the net effect is that you can scan just as low as you can see in all directions and filter out everything which is boring (i.e. the ground doesn't actually move very much, so really it's a fixed background on your radio image).

The other element of this, is AWACS: AWACS radars are higher, and look down. So not only do they go much further, but there is no "below radar" with them - they can be operated from much further away, and have all the same advantages (i.e. much better signal return discrimination). An AWACS will see you on radar long before you're in weapons range (hence the body of them orbiting near the Ukraine border these days).

The final element is that "flying below radar" was also just plain never that effective. You can test this by asking how often it's actually been done, if it's so effective. If it worked all the time, then that's what military's would train to do. Instead the only real advantage it provides is it reduces light of sight, and that's not uniformly applicable - i.e. a radar on elevated terrain would be able to spot aircraft making low approaches around it because it can just look down and pick out the airplane-like returns and ignore the ground scatter - and said radar can be far behind the lines.

I doubt an autopilot would have regained control in a flat spin without an ejection to drastically change the dynamics of the aircraft. I also made the point later that a sufficiently capable AI that could would essentially be able to replace the pilot, rending them moot.

Look, the OP clarified that he meant he wanted an autopilot that tries to lawn dart the plane in an incident of ejection, the purpose of which was to save sensitive technology from the enemy. I still maintain that is silly and if we funded such a program it would be a waste of money.