I hear this sentiment a lot. Having a higher percentage of cars following safe practices is safer even if those are cars aren't perfect and even if 100% of the cars aren't automated.
Can you tell me the logic (I've seen it many times in posts) that there won't be any safety benefits until all cars are automated (or until there aren't any manual drivers?).
I disagree that the benefit will be that discrete.
Why are Google and other companies trying to develop this technology if all it takes is a single fatal crash, and therefore a single lawsuit? It should be in their rational self-interest to avoid any contact with this technology. It would be lighting on money on fire.
This is not the strongest example, but automobile companies have ignored fatality inducing problems, yet the practice didn't end their business. There is likely enough profit in the pursuit to deal with such issues. These lawsuits are slow-moving enough that there would be room to maneuver.
The greatest threat, in my opinion, is over-reacting politicians passing laws due to a shock in public opinion on the safety of the cars. The shock could be legitimate, due to reckless implementation, or just due to an unpreventable accident.
There's also the vaccine industry. Some vaccinations result in severe adverse reactions in a handful of people. But the public health benefit is so vast that the government has stepped in to shield vaccine makers from the lawsuits.
The vaccine point suggests to me that it's not so black and white about if self-driving cars are doomed. If the public safety benefit of self-driving cars becomes so great, then maybe the government will be pressured to ensure they're here to stay. I think there will definitely need to be changes in the law to accommodate this tech, but it's not impossible.
In any case, I expect the incipient googlecars will have capacitive sensors on the steering wheel, requiring that your fingerprints are on the controls.
(that said, I don't think it makes the car itself physically safer, that is up to the manufacturer)
I'll take my chances thanks. Road deaths have been falling for years, despite the number of cars rising. More people die from accidental falls.
As a software developer, there's no way in hell I'd trust a car driven by software.
You're already trusting a car driven by software that could conceivably "be hacked" to hurt you. Algorithms already control things like antilock brakes, shifting, adjusting the suspension and deciding when to deploy the airbags. And assuming you have a cellphone, the NSA is already tracking your every move.
> More people die from accidental falls.
Source? The CDC appears to disagree with you. (my source: http://www.cdc.gov/nchs/fastats/accidental-injury.htm )
Yes, the first driverless cars will have all sorts of blind spots, but they will gradually get better whereas human drivers won't. Software-based drivers can learn from accidents and near-accidents that almost any other software-based driver anywhere has had, whereas human drivers are largely limited to learning from their own vastly smaller range of experience. As a software developer you should be able to predict the eventual outcome: programs will eventually in most circumstances beat even the best human drivers just as they now beat even the best human chess players.
Yes, bugs are a real possibility, but compare the number of miles Google has driven and the number of accidents they've had with the general public, and it compares unbelievably well, especially for such a new technology.
Finally, the NSA doesn't need your car to track you. Between your phone, your license plate and your face, you would have a hard time going off grid.
However, regardless of how bad somebody is riding, when something goes wrong, the rider always loses. Also, something that may seem "dangerous and aggressive" to you may be the best way for a rider to get out of a dangerous situation. I use the acceleration and maneuverability of my bike to escape e.g. being boxed in by large trucks on the freeway. I'm just taking advantage of the escape hatch before something goes catastrophically wrong.
Is that confirmation bias as well?
Software glitches can be tested and fixed. Human errors cannot.
Also, as a cheeky sidenote, do you also refuse to use driverless trains, driverless elevators, driverless escalators and driverless theme park rides? :)
And even that is stretching the facts.
Where's the evidence for this?
perhaps if everyone on the road was in a driverless car, and they all communicated and prevented crashes. But that's not going to happen.