The question is, would that person have got in an accident anyway or did AI create an accident for that person. AI should not be creating accidents, period.
Furthermore, if Musk is as brilliant as people say he is, if he is going to make these claims then he should have a meaningful plan on distributing his vehicles so that enough people own them for actual statistical change. I suspect most people will not like how he thinks this will unfold. It probably calls for an end to private ownership which will erode the freedom for everyone to move around. Considering the tiny chance of getting in more than a fender bender in anyone's lifetime, along with the fact that everyone accepts the risk of the road, I question whether the end outcome will be worth it. It's worth discussing but hardly decided yet.
When autonomous vehicles come out, it is clear that the provider of the autonomy must carry all liability for anything that happens. If it is true autonomy then the person in the car might not even have a valid drivers license.. if there is a person in it at all. Nothing else is logical.
Furthermore, Autopilot only drives the simplest miles of road and always has a human monitoring it. It shuts down if something complicated happens, and it doesn't attempt difficult manoeuvres such as passing other vehicles. So in order to make some sense of the numbers, you must omit all accidents by people not happening within the ability of Autopilot.
These vehicles will only save lives if more than half the drivers are in them. Until Musk presents his plan on how to make his vehicles affordable by 75% of the population, this is all hot air. It must be nice to live in a world where all things somehow get cheap enough for everyone to own one.
So basically automated cars will suck. We will always need a steering wheel and be ready to take over. My biggest concern is that these things will creep along, stopping frequently, basically being frustating for human drivers. This will make traffic problems much worse. Stop and pull over to the side of the road in a snowstorm? Really? What if it is a major through fare with no parking lane, or on the highway and the shoulder on that section is not drivable?
I have old usb keyboards and mice kicking around that we use. Also we use flash drives a lot and android phones. We have some usb mics.
If I have purchased a $2000 computer why should I have to keep a set of dongles around so that I can use the system properly? A $2000 computer should do everything I need it to out of the box. It shouldn't have to be 'completed' with a $5 dongle.
Why are you talking about breaking the law? Where I come from, they don't tell you not to come into work just because there is a snow storm. People MUST drive in these conditions.
I find it very funny how easy some people think the software will be for these vehicles. Really, thinking you have all the hardware you need on the vehicle is not a very big deal. I could strap 22 cameras to my car and I wouldn't have much. So far it looks like Tesla has a start with Autopilot but the functionality is currently very lacking; people I have talked to says that it disengages frequently and still needs clear lane markings. So even if Tesla has the right hardware, the real question is how long will the *software* take to get from where it is to fully driving and that is currently unclear. Tesla has done the 80% easy part, but how long will the hard final 20% take? Elon Musk seems to make a lot of cheap comments like this, that get people exited and veil the real issues at hand.
I don't know if Tesla cars can understand the same visual cues as a human, that's why I'm asking. Using fleet learning would be very risky in the winter unless you teach it how to drive around stuck cars, snowbanks, snow clearing equipment, etc, real well.
It's going to need a heat sensor if its going to differentiate between an animal crossing the road and a bag blowing across the road, unless they are really confident they can educate the car on tracking the motion of each and determining what it is as the car approaches.
Slashdot Mirror
So.... your laptop drives you to work?
I was told in yesterday's Tesla article that driving in a blizzard is illegal in the US.
The question is, would that person have got in an accident anyway or did AI create an accident for that person. AI should not be creating accidents, period.
Furthermore, if Musk is as brilliant as people say he is, if he is going to make these claims then he should have a meaningful plan on distributing his vehicles so that enough people own them for actual statistical change. I suspect most people will not like how he thinks this will unfold. It probably calls for an end to private ownership which will erode the freedom for everyone to move around. Considering the tiny chance of getting in more than a fender bender in anyone's lifetime, along with the fact that everyone accepts the risk of the road, I question whether the end outcome will be worth it. It's worth discussing but hardly decided yet.
When autonomous vehicles come out, it is clear that the provider of the autonomy must carry all liability for anything that happens. If it is true autonomy then the person in the car might not even have a valid drivers license.. if there is a person in it at all. Nothing else is logical.
Furthermore, Autopilot only drives the simplest miles of road and always has a human monitoring it. It shuts down if something complicated happens, and it doesn't attempt difficult manoeuvres such as passing other vehicles. So in order to make some sense of the numbers, you must omit all accidents by people not happening within the ability of Autopilot.
Elevators are completely isolated. Enough said.
The problem is, he presented it as "own my most amazing technology" and not the truth which is, "Please be my beta tester".
"You have just experienced 1 hour and 20 minutes of 100% safe driv--
These vehicles will only save lives if more than half the drivers are in them. Until Musk presents his plan on how to make his vehicles affordable by 75% of the population, this is all hot air. It must be nice to live in a world where all things somehow get cheap enough for everyone to own one.
So the easy human monitored miles that autopilot can navigate are safe? That's simply astonishing!
They don't close the highways as soon as the lane markings are under snow. They close the highways when they are impassable by passenger vehicles.
So basically automated cars will suck. We will always need a steering wheel and be ready to take over. My biggest concern is that these things will creep along, stopping frequently, basically being frustating for human drivers. This will make traffic problems much worse. Stop and pull over to the side of the road in a snowstorm? Really? What if it is a major through fare with no parking lane, or on the highway and the shoulder on that section is not drivable?
So basically
I thought you had to use 'Apple approved' dongles with Apples.
I have old usb keyboards and mice kicking around that we use. Also we use flash drives a lot and android phones. We have some usb mics.
If I have purchased a $2000 computer why should I have to keep a set of dongles around so that I can use the system properly? A $2000 computer should do everything I need it to out of the box. It shouldn't have to be 'completed' with a $5 dongle.
Where I come from it's just a way of life. We don't shut down for every bad snow storm. Automated vehicles will have to work in such weather.
I still don't understand what happens when the route changes, due to construction or other obstacle. Fleet learning will only take you so far.
Ok well I recently saw a comment indicating that Autopilot would turn off if the lane markers were covered, so I found that confusing.
Why are you talking about breaking the law? Where I come from, they don't tell you not to come into work just because there is a snow storm. People MUST drive in these conditions.
I find it very funny how easy some people think the software will be for these vehicles. Really, thinking you have all the hardware you need on the vehicle is not a very big deal. I could strap 22 cameras to my car and I wouldn't have much. So far it looks like Tesla has a start with Autopilot but the functionality is currently very lacking; people I have talked to says that it disengages frequently and still needs clear lane markings. So even if Tesla has the right hardware, the real question is how long will the *software* take to get from where it is to fully driving and that is currently unclear. Tesla has done the 80% easy part, but how long will the hard final 20% take? Elon Musk seems to make a lot of cheap comments like this, that get people exited and veil the real issues at hand.
I don't know if Tesla cars can understand the same visual cues as a human, that's why I'm asking. Using fleet learning would be very risky in the winter unless you teach it how to drive around stuck cars, snowbanks, snow clearing equipment, etc, real well.
It's going to need a heat sensor if its going to differentiate between an animal crossing the road and a bag blowing across the road, unless they are really confident they can educate the car on tracking the motion of each and determining what it is as the car approaches.
Having 'information' is a small part of automated driving. Possibly 2% of the problem.
What about this article makes you think that?