I feel like this rapidly moves into Trolley Problem territory. 10,000+ people die in a year due to drunk driving, what's the acceptable death toll for rolling out driverless cars as fast as possible?
Additionally, if driverless cars are being tested on public roads, you need laws and practices that are robust enough to prevent the kind of accident that killed someone. They had only one operator in the car, who was responsible for monitoring the car and logging information about its performance. They had specifically disabled safety systems which would have caused it to brake sooner.
Even if its expected that driverless cars will reduce fatalities due to drunk or distracted driving, you need to have sufficient evidence of the improvement, safety systems in place for test systems, and public discussion of the tradeoffs and what situations driverless cars might perform worse in than human drivers, because lots of people have behavioral expectations based on human drivers; things like making eye contact with a driver before crossing the street.
So yes, it is good to be working on and testing driverless cars, but you need to have a serious sense of humility and safety about it, and not push it, because even if you are incredibly safe you can expect some backlash, but if you miss the mark at all and oversell the safety, you'll get a lot more backlash, with justification, much sooner, which could set the whole program back significantly.
No. Considering the other benefits of self-driving cars, such as increased mobility for the elderly, you can at most require that they are as safe as cars with drivers, not necessarily safer.
If the benefits were great enough (reduced pollution indirectly saving lives? saved life-years due to more efficient traffic?) one could even accept that they could be a little less safe as far as direct fatalities are concerned.
I disagree. I'm finding it hard to come up with an example that would make me think it's ok to have more people hit and killed by cars. I'm more than ready for self-driving cars to be a reality, but not if more people will die. "Saved life-years" is not good enough.
> I'm finding it hard to come up with an example that would make me think it's ok to have more people hit and killed by cars.
Entirely speculative: Heart failure is the leading cause of death in the US. If 100% self-driving traffic could let EMTs arrive faster at the scene that could save more lives than a slight increase from traffic accidents.
Humans are disproportionately swayed by proximate causes, so deaths directly attributed to self driving cars are going to be more heavily weighted in the court of public opinion.
increased mobility for the elderly
So at this point, self-driving cars still have quite a bit of work to go to prove that they are actually safer than a drunk driver, let alone a sober driver.
[0]: https://www.nhtsa.gov/press-releases/us-dot-announces-2017-r...
Unknown sobriety != sober.