If I recall correctly, the NRSC mask and pre-emphasis were adopted around 1988 or 1989..
Way back in the late 1930s, parallel with the development of FM, there was an "experimental" AM service with stations between 1500 and 1600 kHz that were granted 20 kHz channels. There were four of them - New York City, Connecticut, Kansas City, and Bakersfield, California. They lost this status when NARBA came into effect in 1941 and were assigned standard channels. None of that affected the actual frequency response of the stations; it affected the likelihood of interference from adjacent channels. By then, it was clear that FM was going to be the high-fidelity medium because of its much better ability to reject interference. The Kansas City experimental station, W9XBY, which is the one I know the most about, only survived another year.
It's very difficult to get clean reception of an AM signal, especially in urban areas. There's atmospheric interference that's always been. Plus computing devices are far more widespread than they were just 30 years ago. They generate noise. Electrical infrastructure is aging - a cracked insulator on a 13.8 kv line can spread noise over a wide area - with some utilities having neglected maintenance in order to avoid unpopular rate increases or reductions in actual rates of return. To phrase it another way, once an AM signal leaves its transmitting tower, it's subject to additional electrical chaos that affects the quality of the signal. FM is able to resist that chaos to a much greater degree than AM.
As for AM stereo, my experience is that such stations, at least on an analog receiver, are harder to tune - to hit the center of the station's bandwidth precisely. If not tuned precisely, there's a substantial amount of distortion. To be fair, I've never used a digital AM stereo tuner, so that could solve such problems. But then there's the processing needed to keep the station's loudness at a level where it can drown out much interference. Then there's the receiver's own AGC.
FM has its own issues, the pre-emphasis curve foremost among them. Usually, though, if an FM station sounds like, you know that the station intentionally did that for whatever sick reason it had for doing so. An AM station has much less control over its audio destiny.
But...all this is what mathematicians call "a necessary but not sufficient condition". Programming that enough of an audience will be interested in also has to be part of the equation. Right now, that's the core problem. Audiences have many more choices that are more interesting to them than what's on FM, or AM. Traditional radio is stuck in a rut. It has been for decades, but that was not so evident because there was a lack of alternatives. Now there are plenty of alternatives.