Post

Conversation

In other words, the current safety data may understate the benefits of automated vehicles; as they are rolled out more widely, fewer of these accidents should happen.
Quote
Timothy B. Lee
@binarybits
Not only do Waymos get into fewer crashes than human drivers, the overwhelming majority of those crashes appear to be the fault of another driver.
Image
David Watson 🥑
Post your reply

Doesn’t that depend on whether the drivers who are causing the accidents will choose to drive in automated vehicles in the future? I’d expect ‘bad’ drivers to prefer driving themselves, just as I suspect the high death rates of motorcyclists are as much a correlation to the kind
I think it is broadly true - especially as V2V protocols take hold - that the most dangerous AV scenario is when there is only one AV on the road amid a sea of human drivers (esp. if they are ATL drivers) Vehicle-only crashes go to zero if there are only AVs on the road, right?
Having been in computing for over 30 years, most of that finding other people's software defects, my concern about self-driving cars is that a single bug could result in catastrophe. All software has bugs, and huge complex software has many, many bugs.
The thing is, the places they are currently rolled out in are not a representative sample of the nations roads. I suspect they are selecting for places where best suited for the car's sensors.
I’ve never lost a game of chicken to a Waymo at a 4 way stop-sign. So far they don’t show the guts to go all the way.
So we can again walk on the streets and not fear being ran over, right? We can reclaim the roads for... you know... humans, right?
I would love to see these as an option for people with certain disabilities and the elderly. People who can technically drive, but it would be safest for them to act more as copilot.