ISA is supposed to also work of GPS data, not just road signs. If the implementation is that bad, it calls for change.
At least in my cars, and can myself change settings so that the car doesn't slow down itself for a speed change, and also change the sensitivity of emergency stop.
The idea behind isn't bad and had nothing to do with Greta. Trucks have been speed limited and enforced for ages, imagine them being able to do 160km/h downhill... It's similar for cars, lesser weight (but gap keeps closing...) but much bigger in number, and consequences are by the laws of physics so much worse with the many exponents of speed (as both compound risk and also consequences increases exponentially with speed) and people are very bad in doing proper risk assessments - 3700 people die every day in traffic. Globally but still quite some. For airplane crashes that number is 2. Not 2000, but two. For context, luckily heavily regulated.
But I also see issues, besides the technology not working good enough, do these systems increase other risk behavior like looking more at the phone... Or perhaps people do that anyway (yes they do in older cars for sure... ) and it's good with additional systems? Quite happy with city safety systems while my kids walk to school...
Lets hope the implementation becomes reasonable so it's actually a help that most people accept, and which actually safe drivers can ignore (when nobody else is around...). Like all tech it will probably improve, have to start somewhere?.
I agree with many of your points, but maybe I should elaborate a bit more by my part.
I worked for years in the data analysis and machine learning fields, and, of course, in the AI sector.
By direct experience I know how faulty computer science is by nature (it introduced the "turn off/on" as a very bad universal way to fix things) and the current mainstream in AI (deep learning) is even more faulty and unfixable (see the miserable end of autonomous driving hype).
ADAS are the most simple declination of autonomous driving, and they work with the same technology (mostly), trying to understand what happens outside a car, just by reading 2-D images and pixels, along with some simple data about distances measured by radars.
Faults and misinterpretations are always ready to emerge in daily drive, where unforeseen scenarios are very frequent (not to mention ever-changing weather and light/dark conditions), and the data on which ADAS have been trained prove to be inevitably missing important parts of the universe of possibilities a driver is forced to manage in the daily driving.
In a word, the totally lack the generalisation our brain continuously applies in our living, and they are able to do well only about the exact things they were trained on, and in near-ideal circumstances only.
They can't and won't work well until it's radically changed the way they are trained, and there is still a long time to get there, if we will ever get there one day.
But in the meanwhile, people should stop thinking that computer science is better than us in managing a car, and they should trust these systems as what they actually are: a very mistake-prone helping tool, to be used the less the possible.
Of course ADAS are probably a necessary emergency solution now, in a world where driving a car is considered like an annoying break between phone-browsing sessions.
But maybe instead of accepting as an inevitable "given" the fact that people will be distracted by every thing while driving, we could make some more steps towards a technology that limits the use of the phone to its very necessary minimum functionality, while being into a moving car.
But something tells me that, for many "profitable" and well hidden reasons, it won't ever happen.
So, this is my personal rant against the world of today, I hope you will forgive me someday...
And it's still like that by now.
