"Hyper-aggressive BMW AI takes out 15 pedestrians and causes $250,000 in collateral city-wide damage. Reason cited : danger to owner's career prospects from possible late business lunch arrival. Thankfully the owner is reported to be safe (and punctual)."
I don't know why this is getting downvoted. Sorry, but our most basic human instinct is that of survival. By that I mean my and my family's survival, not yours. I'm not saying that I wouldn't sacrifice myself to save many other people (no one really knows what they would do until they actually have to make that call), but I'm not going to buy something that will sacrifice my wife and child because a computer thinks it's the best outcome.
It seems to me that - socially - we should regard this as the responsibility of the driver rather than the automaker. They chose to drive in that place at that time, they maintain their vehicle, and they have chosen some balance between safety and heroism (either by tweaking a setting or by picking what car they buy). It's the driver's agency at work.
At first, sure. But the goal of autonomous cars is to eventually remove driver agency altogether. What happens when all cars are fully autonomous? Will they collectively weight their passengers by "killability", perhaps based on age, insurance premium, marital status, social media status, etc?
I don't follow. You're proposing there will be driverless cars pursuing their own agenda, as opposed to the agenda of some human or group of humans who have - in some sense - deployed them? That sounds unlikely before the robot uprising.
I don't know about "pursuing their own agenda" but making their own decisions - yes, probably. There isn't much point to autonomous cars which are essentially drones.
Decisions don't create agency. My router makes decisions about what packets to route (and where) and what packets to drop. It does so based on a policy I set and those decisions are an extension of my agency. If I configure my car to prefer saving me to saving 100 schoolchildren (or pick a car with that configuration over the alternative), I'm responsible for that.
I wasn't speaking about what is (or should be) required by law. I was speaking about our social reaction, which certainly has a relationship with law but is not the same thing.
If it's acceptable for a person, in the heat of the moment, to decide to save themselves at a cost to others (once they've considered what other options are available to the best of their ability), then it's acceptable to decide to put in place a system that does the same, assuming it is (at least) comparably capable of considering other options.
If it's heroic for a person to, in the heat of the moment, decide that it's important to save another person even at risk to themselves, then it's also heroic for them to put in place a system that does the same.
Of course, opting out of the whole thing and sticking with a human-driven car is also a choice, and once autonomous cars are sufficiently better than human drivers it may itself become an unacceptable choice.
In the long run, the law should probably catch up to wherever we've arrived at socially and forbid more egregiously unacceptable choices, but I don't see any reason we should start there.
But to sell an autonomous car, you have to survive lawsuits. That's the paramount issue in front of manufacturers. The engineering issues are simple in comparison.
What's acceptable or not will be decided in a court, unfortunately. And like most other traffic-related issues, some default rule (driver in rear at fault etc) will be built over many cases over time.
Unfortunately its unlikely a reasonable rule will prevail. Instead some enforceable rule will win out. Regardless, early manufacturers of autonomous cars are likely to be sued to oblivion in the process.
What's decided in a court depends in large part on social judgements. We should be deciding what's reasonable and convincing others (and being convinced - as a part of that collective deciding), not whining about things that haven't happened yet.
There is always a balance. At some point these types of algorithms will probably be regulated, and that will take some of the pressure off of both owners and manufacturers.
Given the choice I would usually pick me over you, but I would also be ok with buying a car that would choose a tree over a group of kids. This is especially true given that the autonomous car is safer for me overall, even if it takes that one split-second choice away.
In the long term, if all cars are autonomous and communicating, it's reasonable to think that most situations could be handled by carefully coordinated group action so that no one is injured. (Especially in the example given in the article - if only one car has a problem, the others should compensate.)
In the near term, though? For legal reasons it might be necessary to hand control immediately to a human driver to let them make the decision.
Or perhaps far, far more stringent inspection requirements could help mitigate some of this?
If civil-rights-conscious Americans demand the right to manually control their cars, then its possible an unpredicted human operation (abrupt stop/turn) could upset the autonomous grid and create unsolvable problems - cars going too fast to adapt in time.
Either the entire grid would have to slow down enough for any unanticipated human action (which are unlimited in variety), or they would have to 'push the envelope' and accommodate only the likely behaviors.
This could very likely result in facing the issue raised by OP: does the autonomous car(s) sacrifice itself, or hit the human driver while preemptively engaging safety devices for its own driver?
The designers of the autonomous network will have to face this issue immediately. Its not some remote possibility; its part of the very fabric of the network.
This article goes out of it's way to create hand wringing 'what-if' scenarios. Completely ignoring the fact that car accidents would be practically eliminated with autonomous vehicles.
It's perfectly reasonable for a group of cars to all take action with their owners self-preservation in mind first and foremost. If your worried about 'greater total number of deaths', look at human drivers.
I still think it's an interesting question to think about. There's going to be a (long) period of time whereby autonomous vehicles are sharing the road with human controlled vehicles and the elimination of all/most accidents is not likely during that transition.
18 comments
[ 2.5 ms ] story [ 53.3 ms ] thread"Hyper-aggressive BMW AI takes out 15 pedestrians and causes $250,000 in collateral city-wide damage. Reason cited : danger to owner's career prospects from possible late business lunch arrival. Thankfully the owner is reported to be safe (and punctual)."
Autonomous car algorithms must face this issue immediately. Choosing to serve the driver/owner's interest will be a compelling sales point.
If it's acceptable for a person, in the heat of the moment, to decide to save themselves at a cost to others (once they've considered what other options are available to the best of their ability), then it's acceptable to decide to put in place a system that does the same, assuming it is (at least) comparably capable of considering other options.
If it's heroic for a person to, in the heat of the moment, decide that it's important to save another person even at risk to themselves, then it's also heroic for them to put in place a system that does the same.
Of course, opting out of the whole thing and sticking with a human-driven car is also a choice, and once autonomous cars are sufficiently better than human drivers it may itself become an unacceptable choice.
In the long run, the law should probably catch up to wherever we've arrived at socially and forbid more egregiously unacceptable choices, but I don't see any reason we should start there.
What's acceptable or not will be decided in a court, unfortunately. And like most other traffic-related issues, some default rule (driver in rear at fault etc) will be built over many cases over time.
Unfortunately its unlikely a reasonable rule will prevail. Instead some enforceable rule will win out. Regardless, early manufacturers of autonomous cars are likely to be sued to oblivion in the process.
Given the choice I would usually pick me over you, but I would also be ok with buying a car that would choose a tree over a group of kids. This is especially true given that the autonomous car is safer for me overall, even if it takes that one split-second choice away.
In the near term, though? For legal reasons it might be necessary to hand control immediately to a human driver to let them make the decision.
Or perhaps far, far more stringent inspection requirements could help mitigate some of this?
Either the entire grid would have to slow down enough for any unanticipated human action (which are unlimited in variety), or they would have to 'push the envelope' and accommodate only the likely behaviors.
This could very likely result in facing the issue raised by OP: does the autonomous car(s) sacrifice itself, or hit the human driver while preemptively engaging safety devices for its own driver?
The designers of the autonomous network will have to face this issue immediately. Its not some remote possibility; its part of the very fabric of the network.
It's perfectly reasonable for a group of cars to all take action with their owners self-preservation in mind first and foremost. If your worried about 'greater total number of deaths', look at human drivers.