Software design
Lonewolf50, as someone with an IT background, including supervision of programmers, I have to suspect a fault in the software, in its not matching real human behaviour.
I've found programmers design software in a particular way, and if human people behave differently the programmers are resistant to changing the software. A simple example is entering a date. People might enter a date various ways, but programmers often don't like to design the program to accept any valid date. Often programmers will insist people should have to learn how to enter the date a particular way.
I suspect that is part of why the handling of stall warnings is handled badly. If the warning is supposed to be continuous then it should be continuous if the aircraft is detected to be stalled while in the air. Don't assume pilots will figure out that the plane is still stalled and but that the stall warning stopped for some other reason. Don't assume a pilot who did the right thing will figure out that a new stall warning is misleading.
In this thread it has often been said that somehow the pilots seemed to not realize that they were stalled, or perhaps thought they had gone overspeed. With all respect to those who think the pilots had to be to blame, it is not likely that three experienced pilots all made such errors unless the instrumentation and the way it presented information to them misled them.
BTW, I was surprised when people reacted with surprise to the suggestion that the FBW system should have said why it gave control back to the pilots. I understand the reaction was due to a conceptual understanding of the whole point of the FD system. But I suggest perhaps that should be given a second look. It seems like some information about that was given to the pilots, but perhaps it might in some situations be useful if that information was more clearly available.
I won't go into the weather issue. That, after all, is at least in part a judgement call. And many here have noted that pilots have flown through worse conditions many times without crashing. It could only have been at most a contributing factor, not something sufficient to cause the crash.