In my mind, one of the biggest hurdles I've come across for self-driving cars is the moral dilemma of "Kill the Pedestrians or Kill the Passengers", that is the hypothetical scenario in which to avoid a crash and protect the car and passengers, the only available outcome is to swerve the car off the road and onto the pavement (unavoidably ploughing through pedestrians), or the opposite scenario in which a pedestrian suddenly steps out in front of the car and the only way to avoid killing the pedestrian is to swerve the car into oncoming traffic or of the road into a solid object or down an embankment etc. (both these scenarios assume that merely braking won't slow down the car enough to avoid a collision).Fanghawk said:That's not to say there aren't hurdles to overcome with such a system. Driverless cars would have to be efficiently scheduled to meet everyone's needs, and account for variances between short and long-distance road trips. But the good news is Americans aren't against driverless cars on principle - one study suggested 44 percent would consider buying a self-driving car in the next 10 years. Even if we don't go with the taxi route, that's very promising for the environment and our wallets.
In both these hypothetical situations, the computer running the car is going to have to decide which humans' life is worth more and which human it will kill. Is a driverless car which makes decisions who to kill really something we want on our roads?
Who is responsible for the killing... the programmer who programmed the computer program for this situation (and what programmer would happily write a program where they are responsible for who it kills)? Is the company running the fleet of cars responsible for who it's vehicles decide to kill? Would any company undertake such a liability and responsibility? Could the car be deemed responsible on it's own and the death deemed an unaccountable accident?
Would the cars be fitted with an advanced visual recognition system so it could recognise and prioritise individuals on a case by case basis e.g. kill the passenger to avoid a child, but kill an old person to save the passenger, or would the car be programmed to protect it's customers at all costs and assign no value to pedestrians at all?
As infrequent as a scenario like this would be, it's not impossible and it will be interesting how the companies running these fleets of driverless cars (and the governments legislating them) will tackle this dilemma (apart from making the cars out of soft materials in pedestrian safe shapes and making sure they never travel fast enough to kill pedestrians or it's occupants), although unfortunately this question will probably be avoided and only tackled once enough people have been killed to generate sufficient public outrage and awareness to force an inquiry.