Page - (000092) - in Autonomes Fahren - Technische, rechtliche und gesellschaftliche Aspekte

Why Ethics Matters for Autonomous Cars80 4.3 Next steps Notice that the ethical issues discussed in this paper do not depend on technology errors, poor maintenance, improper servicing, security vulnerabilities, or other failings – and all those will occur too. No complex technology we have created has been infallible. Even industries with money directly at stake have not solved this problem. For instance, bank ATMs continue to make headlines when they hemorrhage cash – tens of thousands of dol- lars more than the account holder actually has – because of software glitches alone [2, 10], never mind hacking. And just about every computing device we have created has been hacked or is hackable, including neural implants and military systems [3, 28]. These vulnerabilities and errors certainly can cause harm in the context of autonomous cars, and it would be unethically irresponsible to not safeguard against them where we can. Putting these technology issues aside and even assuming that perfect technology is avail- able, there are still many other safety and ethical questions to worry about, such as the programming issues above. 4.3.1 Broader ethical issues But programming is only one of many areas to reflect upon as society begins to adopt autonomous driving technologies. Assigning legal and moral responsibility for crashes is a popular topic already [1, 14, 20, 22, 49, 51]. Here are a few others, as part of a much longer list of possible questions: Does it matter to ethics if a car is publicly owned, for instance, a city bus or fire truck? The owner of a robot car may reasonably expect that its property “owes allegiance” to the owner and should value his or her life more than anonymous pedestrians and drivers. But a publicly owned automated vehicle might not have that obligation, and this can change moral calculations. Even for privately owned autonomous vehicles, the occupants arguably should bear more or all of the risk, since they are the ones introducing the machine into public spaces in the first place. Do robot cars present an existential threat to the insurance industry? Some believe that ultra-safe cars that can avoid most or all accidents will mean that many insurance compa- nies will go bankrupt, since there would be no or very little risk to insure against [40, 52]. But things could go the other way too: We could see mega-accidents as cars are networked together and vulnerable to wireless hacking – something like the stock market’s “flash crash” in 2010 [5]. What can the insurance industry do to protect itself while not getting in the way of the technology, which holds immense benefits? How susceptible would robot cars be to hacking? So far, just about every computing device we have created has been hacked. If authorities and owners (e. g., rental car compa- ny) are able to remotely take control of a car – which is reportedly under development for law enforcement in the European Union [50] – this offers an easy path for cyber-carjackers. If under attack, whether a hijacking or ordinary break-in, what should the car do: speed