Imagine a situation in which Uber cars are better than human drivers, but not by a lot. Then imagine some hastily thrown-together patch being pushed fleet-wide that ends up causing crashes in some unanticipated situations. You could imagine quite a lot of carnage happening out there until they're all yanked off the roads and rolled back.
Statistics doesn't tell the whole story because it's backwards-looking, not future-looking, and the Uber crashes aren't going to be independent events.
Statistics of accidents with human drivers also include a lot of humans who are banned from the road and sometimes even imprisoned for not driving safely enough; the average human could and should be driving a lot better than those statistics. And in many jurisdictions, there's an expectation that a commercial driving service is much better than the average person permitted to use a car. If the service aims to reduce transport costs to the extent that vastly more journeys are taken before - which Uber certainly does aspire to - then its autonomous vehicles can significantly increase the numbers dying on the roads even if they actually do have a persistent and statistically significant per-mile safety advantage over human-driven cars.
And if I'm a regulator tasked with reducing road deaths, the question I'd be asking wouldn't be "is this car's autonomous mode marginally better than the average driver so Uber can cease paying drivers and hit profitability?", it's "is the human behind the wheel of a car operating in [semi]-autonomous mode no longer preventing a statistically significant number of crashes?" (or "is telemetry suggesting their interventions are actually responsible for more accidents than they prevent?"). Just because a car is a lot safer with a computer behind the wheel than a car with the average person and no computer behind the wheel doesn't mean that permitting autonomous vehicle operators to rely on technology alone is a safety improvement.
I trust statistics but I don't trust bad statistics. There isn't a self driving car operator out there that is currently testing their cars in statistically representative conditions. And unless they are, there isn't a statistical methodology out there that can say they are safer.
Google, for example, avoids expressways and won't go on freeways, hasn't tested their cars in extreme weather, barely ever sees even bad weather, and is electronically limited to 25mph. I'd be willing to bet the majority of drivers have nearly perfect driving records in those conditions too.
> ... and is electronically limited to 25mph. I'd be willing to bet the majority of drivers have nearly perfect driving records in those conditions too.
Not me: I'd fall asleep if I were to be limited to 25mph all the time ;)
Curious about the 25mph assertion. I've driven next to Google self-driving cars in Austin in my neighborhood numerous times and it was definitely traveling faster than 25mph.
Statistics says that it will take many, many millions of miles driven by autonomous vehicles autonomously before we can know how safe they are compared to humans [1].
The notion that if self-driving technology is statistically safer in bulk than humans, then "it wins" is I think highly fallacious (even though it is made here repeatedly and often).
We as humans do not judge whether to do one thing vs another on that basis. E.g. we would ban widespread gun ownership in countries that allow it at present (very large numbers of people would not die as a result); we would mandate a basic set of vaccinations in countries that at present do not do that (same outcome); we would fit the passenger seats facing backwards in commercial aircraft (same outcome). The list is endless but we do none of these things because people prefer to judge the risks in some activity or behavior in the context of their own near term pleasure or convenience.
Heck we could mandate very simple low-tech driving safety measures for human-driven cars such as top speed limits and a block on driving the wrong way down the freeway.
A common misperception is that we could ban objects, or mandate behaviours. We can only impose penalties on the possession of objects, or mandate punishments on certain behaviours. Simply passing laws does not save lives in the short run.
The one problem I have with self driving cars is that the software is a single point of failure, so one bad patch (one "mistake") could cause millions of crashes
Is there any proof that this is standard for Car companies?
Tech companies are bound to have good procedures for tech... that's what they do.
Car companies? They should be good at car stuff... but... tech stuff?
Look at all the angst surrounding the IOT - companies making buggy/hack-able stuff that otherwise make good things - IE Washing Machines, Refrigerators, etc... look at all the stories about how hack-able stuff involving cars is...
Personally? I don't have a lot of faith in "Old" companies getting tech right.
> They should be good at car stuff... but... tech stuff?
Cars are a type of technology too. I don't think there's any company that's good at "technology". Maybe a university would count? Or one of those big appliance manufacturing corporations.
If you're talking specifically about computers, software, and electronics, I think IT seems more accurate.
You build those statistics by putting them both on the road and comparing the results. Personally I'm not willing to put a machine on the road until it's murdered enough people to properly tally up
Statistics doesn't tell the whole story because it's backwards-looking, not future-looking, and the Uber crashes aren't going to be independent events.