To repeat myself from the other post where I’ll probably be downroaded:
The car should be programmed to self-destruct or take out the passengers always. This is the only way it can counter its self-serving bias or conflict of interests. The bonus is that there are fewer deadly machines on the face of the planet and fewer people interested in collateral damage.
Teaching robots to do “collateral damage” would be an excellent path to the Terminator universe.
Make this upfront and clear for all users of these “robotaxis”.
Now the moral conflict becomes very clear: profit vs life. Choose.
self-destruct
What?
take out the passengers
WHAT?!?
The car should be programmed to self-destruct or take out the passengers always.
Users here don’t understanding the dilemma nor the programmatic aspects.
The car has to be programmed to solve the dilemma on the spot:
- Crush the people outside to save the people inside.
- Intentionally crash into a large object or veer off road and risk crashing into a ditch.
Not talking about it won’t make this go away. It will simply be some decision made by developers and maybe there’s a toggle for the car owner, a kill switch. Either way, it’s lose-lose.
As we’re in fuck cars, I’m assuming that people understand that fuck cars. Why should this impunity of killing with cars be furthered by encoding it in automatic programming? Let the owners of vehicles face the immediate consequences of owning such vehicles. That’s fair. Don’t want to die in your robocar? Fine, drive very slowly and very rarely.
- Crush the people outside to save the people inside.
- Intentionally crash into a large object or veer off road and risk crashing into a ditch.
What?
That’s not what happened here, and I struggle to imagine any situation where that’s the only two options.
I struggle to imagine any situation where that’s the only two options.
Alright, I’ll take this in good faith. Here’s how that happens:
Speeding.
As we all know here, speeding makes crashes way worse, and it makes the braking function fail proportionally.
So, imagine:
The killer road bot is speeding through a street. It’s a bit narrow, there are cars parked illegally on the sides.
The killer road bot enters an intersection and makes a left turn with speed and a there’s someone on a crosswalk.
The killer road bot controls at least these aspects of the car: brakes, acceleration, steering. The brakes can be engaged, but the speed makes them useless in preventing running over the person on the crosswalk. The acceleration is not useful. Everything is happening too fast really, and the killer road bot can’t even calculate which direction the person is walking in on the crosswalk.
The only useful control left is direction by steering. The killer road bot thus has these choices:
- Maintain course, run over person on crosswalk
- Change course
Choice 1 leads to the obvious outcome.
Choice 2 branches out:
2.1. Turn left
2.2. Turn right
If the killer road bot turns left (2.1), it flips the car over and sends it rolling into other cars, thus endangering the passenger(s).
If the killed road bot turns right (2.2), it hits a large tree.
These are the only options.
edit: typo
I mean a autonomous vehicle should be programmed to not speed and even not drive faster than reasonable in the present condition.
In switzerland we have a law that you are not allowed to drive faster than the speed with wich you can come to a full stop at the farthest spot on the road that you can see. (So in a curve you have to drive slowly, because there could be something on the street right in front of you.)
If a autonomous vehicle respects such rules, then it at least has eniugh time to calculate several outcomes and choose one which has the least damage potential.
The trolleyproblem is not applicable here as its not a theoretical situation but a practical one.
