Those numbers include pedestrian deaths, which the same wiki article states "began rising in 2010, and exceeded 6,000 by 2018" (constituting a higher percentage of deaths in 2018 than in 2010).
If you subtract off the 6400 pedestrian deaths in 2018 and the 4200 in 2010, then you do see a drop-off in fatalities per hundred million miles driven -- from 0.971 in 2010 to 0.935 in 2018 (and even lower in 2019 at 0.925, as the raw number of deaths went down as miles traveled went up).
(Further, this isn't even considering other non-automobile occupants in crashes, whether that's bicycles or motorcycles.)
The car isn't even that long--the back pillars and back window are just weird (Hyundai Veloster).
Which brings me to the point of common sense. Not everyone shares it, or uses it, and so these features are useful. We can talk all day about what people should or could do, but at the end of the day, it just isn't going to happen.
It's not clear to me how much of this change to the car shape is driven by competing safety standards, i.e. side impact and rollover protection, and how much it is the continuous march of fashion/stylistic tweaks.
I was in a near crash without doing any of those things (night driving on a very fast road in an area I was unfamiliar with, and misread the shape of the road). I panicked and swerved without checking my blind spot (it was an empty road, or so I thought) and almost caused the other person to crash. More advanced blind spot monitoring systems could have largely removed the danger of the situation I was in, although obviously there is a lot that I could/should have done better myself. I think the next year's model had a HUD which showed blind spot status.
So your question should be, across all cars, how many accidents have there been with and without blind spot monitors? Even then, it would be hard to control for all other factors (newer cars have blind spot monitoring, but are also safer in general, you need to compare similar years, traffic conditions, etc).
All of that to say — any single comment online is just an anecdote.
Which is to say "quite meaningful", but when the absolute rate of serious crashes per driver is as low as it is, it's statistically impossible that driver aids are saving the typical individual driver from a serious crash multiple times in a driving lifetime, let alone multiple times since their introduction.
Maybe new features like self-driving aren't as interesting, but I would buy a new car instead of a used one just for the perceived improved safety.
I don't know if there's any research showing that these features actually reduce accidents/fatalities though (plausibly if they malfunctioned it could be worse than nothing)