Great, now we have cars deciding whether pedestrians are expendible or not...
https://www.theverge.com/2018/5/7/17327682/uber-self-driving-car-decision-kill-swerve
If you don't have an account you can sign up for one now.
You can also reset your password if you've forgotten it.
News, media buzz around the industry, cool images, and more. Share them here, and discuss with others in the rideshare industry, including Uber drivers, passengers, journalists, and rideshare companies.
Great, now we have cars deciding whether pedestrians are expendible or not...
https://www.theverge.com/2018/5/7/17327682/uber-self-driving-car-decision-kill-swerve
Comments
It sounds like their software was getting a lot of false positives on pedestrian sightings, so rather than make its detection software better, they decided to just raise the threshold of what the car considers a pedestrian.
So... who is liable here? This sure seems like criminal negligence if you ask me...
What the heck? I mean, what I want to know is, what sort of things were triggering these false positives? Why were they so confident that changing the threshold could be safe?
I read the article and got the same thing, but it's such a foolish decision that I can't believe anyone actually made it. Did they forget real lives are at stake???
apparently. I am begining to despise this world we live in.
Maybe it would have run into a pack of children if it swerved to miss that lady. Duh. You weren't there.
Well, there was no pack of children, but this does bring up a really interesting (and scary) aspect of this technology... how does a car value different people's lives?
Imagine a situation where a child runs out in front of a car. The car sees the child, and has time to swerve to avoid the child, but the only direction it can swerve is onto the sidewalk, and walking down the sidewalk is an 80-year old man. The car has to choose between killing a child or an 80-year old man. What does it do?
A human driver in that situation would react in some way and hit either one of them, but we'd chalk it up to a terrible tragedy. Humans can't react rationally in such a short time, so it is hard to see the human driver as anything but another victim of a terrible circumstance.
What about the car though? Computers can process information like this virtually instantaneously. The car has time to think. And it has to make a decision. But cars are programmed by humans, which means that a human at som…
Read more...
Well, there was no pack of children, but this does bring up a really interesting (and scary) aspect of this technology... how does a car value different people's lives?
Imagine a situation where a child runs out in front of a car. The car sees the child, and has time to swerve to avoid the child, but the only direction it can swerve is onto the sidewalk, and walking down the sidewalk is an 80-year old man. The car has to choose between killing a child or an 80-year old man. What does it do?
A human driver in that situation would react in some way and hit either one of them, but we'd chalk it up to a terrible tragedy. Humans can't react rationally in such a short time, so it is hard to see the human driver as anything but another victim of a terrible circumstance.
What about the car though? Computers can process information like this virtually instantaneously. The car has time to think. And it has to make a decision. But cars are programmed by humans, which means that a human at some point has to decide what the car is going to do when it runs into this situation.
What does it do? How are AV companies dealing with these potential situations today?
Read less...
Well, duh. That's easy. You run over the 80-year grandpa. He's lived a long fruitful life. Amen.
Here's another one. What if the car is driving across a bridge. A kid runs off the sidewalk in front of the car. The car can swerve but will fly off the bridge, killing its occupants. What does it do?
Are these cars programmed to protect their occupants at all costs?
What if the car could also choose to swerve into oncoming traffic, ensuring a head-on collision. But the car calculates that there is a 50% chance that everyone will survive, but heavily injured. What does it do?
Re: Protect the occupants vs. others.
As with any invention, marketability will come into play at some point. I spoke to a designer on this, off the record, and he told me that they have to design it to protect the occupants or such software will not sell or be adopted. People don't want to buy or invest in a car that may turn on you.
Of course, the landscape may change if this is the government or a large fleet owner (Waymo, Google, etc.). They may be able to decide this for the cause of greater good.
But what is "greater good"? Who decides that? Does the government come up with a set of specifications for whose lives are more valuable? A big ordered list? What are the classes of people on that list? Do government officials get a bump up the list? Do homeless people get shoved to the bottom? Do special interests get to lobby for higher placement on the list?
These are rhetorical questions of course. I just think it's a fascinating thought experiment -- and one that may become real in the not-so-distant future.
BTW, this % survivability and prioritizing human life. It's like teaching the cars to have a conscience and soul.
Maybe. But I don't think so. This is about a programmer back in the lab deciding whose life is more important. These cars are nothing close to AI... yet.
need more details. There is more to the story here taht they are not telling us
Yeah so this is why we have regulations, people.
Soon people will begin feeling bad for the car. "The car didn't mean to hit that person" "Don't hurt the car"
Isn't there a star trek episode on determing if a robot has feelings or something?
I think you are talking about the Episode "Measure of a Man".
Just read an article where they want to try this in PA. This time they are using 2 drivers. How is that saving or making money for Uber? Isnt the whole idea to have self driving cars? Omit the driver?
You know you have hit the future when people start describing robots with human characteristics...
This whole thing needs to stop. it is mind blowing that anyone thinks having robots driving cars around on our streets is a good idea. do we need a better example of why?
Yeah, but people kinda suck though.