There are tons of modes in this data. Assessing all crash fatalities as some sort of "general" problem you can target misses this fact.
The single greatest thing you could do to reduce fatalities is to eliminate drunk and impaired driving.
This discussion is about _mandating_ AEB, not it's mere existence or implementation, and of the overall value of that proposition; which, I am suggesting is low relative to all other problems and efforts spent mandating this would be better spent elsewhere.
> If anything, AEB would help mitigate some of the impact from the drunk and impaired driving.
Are these systems tested for this use case? I'd like to see that. Otherwise, this is a guess, and a bad one.
But this is not mutually exclusive. For one thing, we can do two things at once. For another, automobile fatalities typically have multiple contributing causes. If we had stronger safety regulations around vehicle height, stopping distance, windshield visibility, etc then some number of the crashes involving alcohol or drug use might have been injuries or close calls instead of fatalities.