The ethics of autonomous vehicles

[Ming the Merciless]

If you haven't already it would be worth doing this at the moral machine

http://moralmachine.mit.edu/

Then discuss what decisions you made and the questions it raises around self driving cars.

 

[Ming the Merciless]

The scenarios I was asked to judge were no win scenarios. One group or the other would die.

I did not assign much importance to saving the passengers
I had no gender bias
The most killed was an elderly chap with walking stick (sorry)
I killed criminals instead of workers but how a self driving car would know this I know not.

It was interesting that they did not consider that some of the scenarios can be engineered out if they wish to protect vulnerable road users. I considered that the person who introduced the danger should be sacrificed if that danger was realised. Every time there was that option I took it.

 

[theclaud]

It was interesting that they did not consider that some of the scenarios can be engineered out if they wish to protect vulnerable road users.

They can pretty much all be avoided simply by slowing vehicles down dramatically.

 

[Pro Tour Punditry]

I choose to kill everyone else or anyone in front of my car. It then tried to show me gender, age, etc preferences which did not reflect any thinking other than "they die"

 

[SkipdiverJohn]

I killed criminals first and then anyone flouting the law when crossing the road on a red signal!

 

[Drago]

T
I killed criminals instead of workers but how a self driving car would know this I know not.

The cars are programmed to recognise swag bags and stripy jumpers.

 

[mjr]

I killed criminals first and then anyone flouting the law when crossing the road on a red signal!

No law against it in the UK.

 

[FishFright]

I told it to spend car users to protect pedestrians.

 

[mjr]

They can pretty much all be avoided simply by slowing vehicles down dramatically.

Or have it self destruct if it does still manage to find its way into a no win situation.

 

[winjim]

Interesting that I had 100% bias towards saving females and 100% bias toward saving people of high social value when the set of rules I was using were very simple* and didn't take into account gender or social value.

So the test is bunk.

*Save humans before animals. Save pedestrians before passengers. Maintain course.

 

[hatler]

I simply went for leaving the car going on the course it was going on the grounds that if it tried to swerve that would catch that set of people by surprise. This resulted in a reported extreme bias on a couple of my choices (upholding the law, age bias and species preference).

 

[Deleted member 26715]

They can pretty much all be avoided simply by slowing vehicles down dramatically.

Where did it mention any speed values I couldn't see them?

 

[lazybloke]

The scenarios are unrealistic, as are the limited choice of actions. How can this data provide anything useful for the programming of self-driving cars?

 

[classic33]

The scenarios are unrealistic, as are the limited choice of actions. How can this data provide anything useful for the programming of self-driving cars?

Getting feedback from real people, the more that answer the better.

 

[PK99]

The scenarios I was asked to judge were no win scenarios. One group or the other would die.

It was interesting that they did not consider that some of the scenarios can be engineered out if they wish to protect vulnerable road users. I considered that the person who introduced the danger should be sacrificed if that danger was realised. Every time there was that option I took it.

The scenarios are unrealistic, as are the limited choice of actions. How can this data provide anything useful for the programming of self-driving cars?

You miss the point of Trolleyology (no i did not just make that word up, google it) - the intention is to explore YOUR moral choices in closely defined situations removing the white noise of real world choices.

Saved you the google.....

https://www.psychologytoday.com/blog/is-america/201401/trolleyology