Is there a perfect solution to the trolley problem?

[DaveC426913]

I've never seen a perfect solution to the trolley problem that lives entirely within the given parameters of the problem (i.e. does not introduce any components or actions that are not explicitly stated in the problem).

 

[Melbourne Guy]

If only all thorny problems like this were so easily sidestepped, @DaveC426913

More seriously, I find the trolley problem genuinely interesting as applied to autonomous vehicles because the context ultimately seems to fall back to a cynical, "Who are we going to sue?" It is easy to hold a human driver accountable. Some disembodied machine intelligence? Not so much!

 

[DaveC426913]

More seriously, I find the trolley problem genuinely interesting as applied to autonomous vehicles because the context ultimately seems to fall back to a cynical, "Who are we going to sue?"

This has been batted around a lot.

I'd say the general consensus is it's a pretty contrived scenario where an AI is forced to choose. Simply braking mitigates almost all realistic scenarios.

 

[Melbourne Guy]

I'd say the general consensus is it's a pretty contrived scenario where an AI is forced to choose. Simply braking mitigates almost all realistic scenarios.

Absolutely, @DaveC426913, and a lot of the AI-oriented ethical outrage fails to acknowledge that human drivers are generally selfish when it comes to accidents. We'll instinctually swerve away from another vehicle, putting our passengers in the line of fire. We'll work our way through this stage of resistance / adoption and at some point the young 'uns will look back and wonder at what all the fuss was about.