The taxi comment was funny to me, because here in Germany our taxis are all very clean and well kept and usually also moderately expensive cars like Mercedes.
One thing he failed to mention here is that all software fails eventually. So imagine one of those fancy graphics scenes he commissioned being a 100+ car pileup because a software glitch caused a car to make a sudden right-hand turn to a street that doesn’t exist.
He did mention it. Even showed an example with Tesla causing a 8 car pile up
He showed an example of that happening today. He seemed to be of the impression that something like that can eventually be ironed out, which it can’t. At least that was my interpretation.
Hmmm, for something that took so long to make, with such a visibly huge production value, I think the point is rather weak.
The video spends a lot of time on “but AVs cause accidents too”, without ever opening a single page of statistics, just some horrible articles for shock value. Those few examples are then followed by “drivers kill 40k a year”, which gives me entirely the wrong impression.
Then it veers off into plausible-sounding, but ultimate unsupported wild speculation, with only a single example about a town going carcentric getting ruined (I’d love a case study on that, it seemed far more interesting than most of the video).
For a channel that’s normally very good about showing examples and supporting data i can’t help but notice the stark contrast in this video. And that’s not a good look, especially since the case could be made without proclaiming cybernet to be the end of walking.
The video spends a lot of time on “but AVs cause accidents too”, without ever opening a single page of statistics, just some horrible articles for shock value.
What statistics? Self driving cars are, at this point, barely even a thing - and only in very specific places. Any sort of statistics on that matter, if they exist, would be heavily skewed due to the low numbers.
I’ll add that such statistics are very much a moving target, since AVs are still “getting better every day”. The software is (and will be) under constant development, and there will likely be tradeoffs between safety for pedestrians and convenience for passagers (e.g. how sensitive is the trigger for an emergency break?)
Looking at it as an ongoing relationship between AV operators, regulators and people makes a lot of sense to me. I agree with the points of the video, that operators will likely push for a “just safe enough” standard and try to offload responsibilities onto bystanders.
There’s also a common argument that the problem in AV accidents is primarily the other human drivers, which is a classic case of “if everyone just immediately changed over to doing things this way it would solve the problem!”
His overall point appears to be that a city fully optimized for self-driving cars would be a hellscape at ground level, even allowing for fewer accidents, so no real reason to belabor that point, which is mostly made in service to pointing out how dumb it is when your solution to reducing accident rates is “buy a new car” instead of anything systemic. like improving mass transit.
You want studies about a potential future? First I don’t think that’s reasonable and second those would be funded and poisoned by the car lobby. Massive industries have been capturing and derailing scientific research since asbestos and radium.
I totally agree with the author learning from history is our best approach.
