[masraum]

Originally Posted by Tishabet

I spent two years in AI and autonomous driving for a GM-backed startup, long story short I have expertise with these and have lots of experience riding in autonomous vehicles both on the track and in major cities globally. Take it from the guy who has seen all of the data: level 4 autonomous vehicles (e.g. Waymo) are demonstrably safer drivers than humans and have been since around 2021.

But, safety and being a better driver than a human is not the entire picture... there are a lot of legitimate complaints about robotoxis. One of the biggest real world complaints about them (and also about level 2 autonomy like you have with Tesla "self driving") is that the vehicles follow the law to a fault and that is actually pretty annoying to human drivers, myself included. They drive the speed limit, they will never cross a solid white line, they will never act assertively to someone jaywalking etc, and by being such great followers of the law and rules of the road they are annoying to many drivers. You can tell your human taxi driver to "drop me off right here" and the driver will pull over illegally, you see that all the time and don't think twice about it... but a robotaxi won't do that because that is breaking the rules.

This thread and some of the responses therein highlight some of the real world headwinds to autonomous driving. Being demonstrably better and safer drivers than humans is not sufficient for many people to accept robotaxis or level 4 autonomous vehicles generally. Wrapped up in this is a simple truth: if your loved one is hit and killed by a human driver that is a tragedy, but a tragedy that is somehow more understandable and ultimately acceptable than if your loved one is hit and killed by an autonomous machine. The result is the same, but something about a machine causing someone's injury or death provokes an outrage which just isn't present with a human driver.

Thank you for responding with so much real world and behind the scenes info and insight. About the last point, you are absolutely spot-on. If a human gets in an accident, that sucks, but doesn't seem crazy. If a machine gets in an accident, then that seems incomprehensible because a machine is not expected to make a mistake despite the fact that the machine is told how to work and what to expect by people that make mistakes and in a complex environment where not every possible scenario can be expected/anticipated (especially when humans are involved).