[superkuh]

Self-driving cars are a bad example for your argument because self-driving cars really are very restricted in their regions of operation. Yes, self-driving cars can work in Arizona where everything is flat, dry, and unobscured. But they just don't work in most parts of the country for half the year because of snow covering markings, emergent lane flocking that has nothing to do with actual lanes, and other winter driving fun. Autonomous cars won't work in these vast regions of the world until they can do the wrong thing the same way humans do.

The hype surrounding them is, and always has been, hype. No substance.

[martythemaniak]

I think you might be spending too much time in echo chambers. The hype probably underestimates the impact the tech will have (it is near impossible to predict higher order effects) and there's been an incredible amount of progress over the last few years.

[superkuh]

The only social media I use is HN and IRC. The only chamber I need to yell into is right outside my door, full of snow with the roads covered. Any car using absolute positioning of self and road will drive in the wrong lane. Why? Because they humans aren't using the right lanes and what the humans are now flocking to as a lane is now the correct lane. And it's not like the sides of the roads remain absolute either. This is not some hour long transient situation. It's for 5 months at a time. Unless an autonomous car can follow the emergent pseudo-lanes (there's no real distinct visual markers either) and make the wrong decision like all the other humans it's going to hurt and kill people.

And that's why you only hear about autonomous car tests in arid warm regions. All the ones in cold climates so far have failed or used special roads with embedded sensors, or a car with many TB of pre-recorded ground penetrating radar scans mapped to road positions. And those absolute positioning based navigation systems would be deadly if mixed with emergent human lane behavior in winter.

I can understand why people who've never driven over a real winter can think autonomous cars are ready. But they'll have to achieve near human emulation of purely vision based driving to make the same errors humans do in winter. And that's a long, long way away.

[martythemaniak]

I'm from Canada, so I'm somewhat aware of winter, yes.

Technology will always arrive at different times at different places. Your area probably got high speed internet and electricity later than cities, and it'll get AEVs later as well. As any technology, it'll target areas with highest benefit:cost ratios first, then incrementally expand. Nothing particularly notable about that and when thinking about the overall development, a 5-10 year difference between tech being available in Toronto vs Yukon is not relevant. I mean, it sucks for people wanting the tech, but that's about it.

Driving in pseudo-lanes is probably much easier than what Tesla's FSD can already do (detect, keep track of, and plan around, dozens of cars and pedestrians), the reason you'll see it last is less tech challenge, but that getting it to work in a city is much more valuable.