If a Google Car Has to Kill Someone, Who Should it Be?

[WanderingGaze22]

Many of us are familiar with The trolley problem, a thought experiment about a fictional scenario in which there are only two choices: save a group of people in danger of being hit by a trolley or divert the trolley to save just one person. Throughout the years there are many versions with different details such as the one person being someone you know or the group being children.

With self-driving cars now being a reality and becoming more and more public, how are the processors taking into account the outside variables? Should a driverless car lunge into another lane in order to avoid a child who just ran into the street? Should it do a full-quick stop to avoid hitting a galloping deer knowing there is a speeding car right behind it? Do these decisions change if the driverless vehicle happens to be a prison bus transporting inmates, or perhaps an ambulance with a pregnant woman minutes away from giving birth? If someone is killed or injured in these scenarios, who should be held accountable? With these vehicles logging more and more hours on highways and roads, check the link below to have the info on where we can go five years after self-driving cars are mass produced.

https://www.pcmag.com/opinions/the-dile ... iving-cars

[Pattern-chaser]

These questions are insoluble, in the sense of there being - or not being - one answer that is universally agreed upon. All we can do, I think, is to play the odds. I believe that tests show that, overall, driverless cars have fewer accidents than human-driven vehicles do.

But the question in the OP title - "If a Google Car Has to Kill Someone, Who Should it Be?" - is difficult, if not impossible, to answer, IMO. Just as the trolley question is unanswerable, also IMO.

[Sculptor1]

Dimitri Dolgov

[Ecurb]

The question is purely hypothetical. If the Google car has a choice between hitting another car and a pedestrian, it should hit the other car, because the accident is less likely to be fatal. The Google Car should be programmed to select those accidents with the least risk to life and limb (if such a choice arises). Hitting the deer (for example) might be more dangerous to humans than slamming on the brakes.

[LuckyR]

Many of us are familiar with The trolley problem, a thought experiment about a fictional scenario in which there are only two choices: save a group of people in danger of being hit by a trolley or divert the trolley to save just one person. Throughout the years there are many versions with different details such as the one person being someone you know or the group being children.

With self-driving cars now being a reality and becoming more and more public, how are the processors taking into account the outside variables? Should a driverless car lunge into another lane in order to avoid a child who just ran into the street? Should it do a full-quick stop to avoid hitting a galloping deer knowing there is a speeding car right behind it? Do these decisions change if the driverless vehicle happens to be a prison bus transporting inmates, or perhaps an ambulance with a pregnant woman minutes away from giving birth? If someone is killed or injured in these scenarios, who should be held accountable? With these vehicles logging more and more hours on highways and roads, check the link below to have the info on where we can go five years after self-driving cars are mass produced.

https://www.pcmag.com/opinions/the-dile ... iving-cars

The cars will be programmed in the way that Google lawyers can best defend against lawsuits from fatalities involving their cars.

[Sculptor1]

The question is purely hypothetical. If the Google car has a choice between hitting another car and a pedestrian, it should hit the other car, because the accident is less likely to be fatal. The Google Car should be programmed to select those accidents with the least risk to life and limb (if such a choice arises). Hitting the deer (for example) might be more dangerous to humans than slamming on the brakes.

That is very easy to say, but not so easy to do.
Driverless cars have already killed.

[WanderingGaze22]

The question is purely hypothetical. If the Google car has a choice between hitting another car and a pedestrian, it should hit the other car, because the accident is less likely to be fatal. The Google Car should be programmed to select those accidents with the least risk to life and limb (if such a choice arises). Hitting the deer (for example) might be more dangerous to humans than slamming on the brakes.

One would imagine that is the way the algorithm will be made. Precisely what I was thinking how the safety features will function.

[WanderingGaze22]

But the question in the OP title - "If a Google Car Has to Kill Someone, Who Should it Be?" - is difficult, if not impossible, to answer, IMO. Just as the trolley question is unanswerable, also IMO.
[/quote]

You may be right about that part still, I doubt it will hinder people from trying to defy the odds altogether. As for the driverless car incident statistics, hopefully the data processors will not be based solely on Intel Core software in order to prove the tech can be as aware of their surroundings as we can be.

[WanderingGaze22]

With self-driving cars now being a reality and becoming more and more public, how are the processors taking into account the outside variables? Should a driverless car lunge into another lane in order to avoid a child who just ran into the street? Should it do a full-quick stop to avoid hitting a galloping deer knowing there is a speeding car right behind it? Do these decisions change if the driverless vehicle happens to be a prison bus transporting inmates, or perhaps an ambulance with a pregnant woman minutes away from giving birth? If someone is killed or injured in these scenarios, who should be held accountable? With these vehicles logging more and more hours on highways and roads, check the link below to have the info on where we can go five years after self-driving cars are mass produced.

https://www.pcmag.com/opinions/the-dile ... iving-cars

The cars will be programmed in the way that Google lawyers can best defend against lawsuits from fatalities involving their cars.

I will not dare to speak for everyone here, but I for one will definitely have issues of a high caliber if this is how the lawyers as well as whoever is going to be defendants are going to behave.

[Memaw18]

Should be no one?

[AverageBozo]

If there’s a “non-driver” in the car, that’s your prime candidate.

If the car is completely driverless, the one who dispatched it should be the one to go.

[LuckyR]

With self-driving cars now being a reality and becoming more and more public, how are the processors taking into account the outside variables? Should a driverless car lunge into another lane in order to avoid a child who just ran into the street? Should it do a full-quick stop to avoid hitting a galloping deer knowing there is a speeding car right behind it? Do these decisions change if the driverless vehicle happens to be a prison bus transporting inmates, or perhaps an ambulance with a pregnant woman minutes away from giving birth? If someone is killed or injured in these scenarios, who should be held accountable? With these vehicles logging more and more hours on highways and roads, check the link below to have the info on where we can go five years after self-driving cars are mass produced.

https://www.pcmag.com/opinions/the-dile ... iving-cars

The cars will be programmed in the way that Google lawyers can best defend against lawsuits from fatalities involving their cars.

I will not dare to speak for everyone here, but I for one will definitely have issues of a high caliber if this is how the lawyers as well as whoever is going to be defendants are going to behave.

I know it sounds callous, but if you are a for profit company and someone has to get hurt, a decision has to be made regardless. It is completely logical to use legal defendability as the criteria. Mainly because such a choice probably would use least overall harm as it's decision point.

Two birds, one stone.

[figliar0]

It is sure that the programmers will decide this question for us. And as one can observe, opinions about morality are changing really fast nowadays... And these changes will be reflected to such a programs. Maybe one will be surprised how will those algorithm work in few years.

That is a big problem of nowadays. Computers are everywhere, managing our lives. But those are not some machines - it is the programmer's code which (almost always beyond our control and knowledge) is executed to perform tasks (sometimes) above our lives. So maybe google's programmer decided in totally different manner as you would (if you are the driver). Question is - are you sure that all decisions google's programmer did are for your sake?

[AwkwardPanda]

Maybe the car will require a "morality customization" program. It presents a series of questions and choices to the owner of the car, about what the car should do in different moral scenarios. The car will not be able to drive before the owner selects their moral preferences. That way, when a self-driving car ends up killing someone, the owner of the car will be persecuted as they would now. It would give the company decent deniability to the crime, and it would allow the owner of the car to think over such moral scenarios, instead of choosing on the spot.

[LuckyR]

Maybe the car will require a "morality customization" program. It presents a series of questions and choices to the owner of the car, about what the car should do in different moral scenarios. The car will not be able to drive before the owner selects their moral preferences. That way, when a self-driving car ends up killing someone, the owner of the car will be persecuted as they would now. It would give the company decent deniability to the crime, and it would allow the owner of the car to think over such moral scenarios, instead of choosing on the spot.

Well if the corporation has an option called "Death Race 2000" for the driver to select who do you thimk is going to get sued, Joe Citizen with a personal wealth of $50,000 or Google?