We don't have 100% autonomous trains yet, yet we're supposed to have self driving cars next year?!

wejo wrote:

He said it wouldn't happen in his lifetime. He's looked to be in his 60s. We didn't talk very long but he made it sound like the .0001% conditions would be the problem. Rain, ice, some kid runs out in font of car, etc.

I think you're holding self-driving cars to a different standard than human operated cars. Human operators struggle under the same conditions that you mention. Why is a self-driving car untenable because it might hit a kid who runs out into the street when human drivers regularly run over people?

self driving cars are the worst idea ever. creating combustion and propelling a hunk of metal at the slight touch of your foot is not enough technology. we now have to want it to be done automatically.

We want the car to be our eyes, ears, and mind. And why? Laziness? Because we could be "doing something better with our time"? dumbest argument ever. Really? That's the reason to risk human lives?

They are no where close to proving that self-driving cars are safer than human drivers. I've worked at a software company that made "sophisticated" computer vision, i.e. computers trying to identify objects in images, and guess what? No one in that company would trust their life with that software, even though it was "best in class" for the task that we created it for.

There are too many variables. Of course the cocky engineers will challenge that notion and say, "oh yeah? watch!" but again, the entire purpose of creating self-driving cars is stupid. Yes, stupid.

There was an article recently posted that said that the engineers of self-driving cars made the decision to prioritize lives outside of the car over lives inside the car. So lets say, you're driving along a coastal highway, sending emails on your laptop on your way to work, because of course, you're taking advantage of all of this free time, and there is a child crossing the road...due to your speed, there is no time to safely brake, so the algorithm decides it's only course of action is to swerve off the road, over the cliff, and you die.

I hope those emails were important.

Ok ignore the kid in the street thing but if the car can not be used when it rains, it won't work as a solution for a lot of people.

The premise is a) they could be safer than humans driven cars and

b) would free up a lot of our time and make transportation more efficient.

As for a) Humans text and drive, distracted drive, drunk drive, make errors in judgement etc, so I can see how a computer driven car doesn't have these problems and could be safer.

The question is when do we get to self-driven cars being safer than humans? You and the guy I talked to last year don't think that is anywhere near. Elon Musk thinks it's very soon. If it's so soon why do trains have engineers?

1. Talk to railroad people. And while you're at it, ask the truck people why trucks aren't diesel electric, like every train out there.

2. ABS.

3. Yes, we're all smarter than the people out there who have companies like SpaceX and Google. That's why we waste time here, and don't create stuff like rockets and Android.

wejo wrote:

Ok ignore the kid in the street thing but if the car can not be used when it rains, it won't work as a solution for a lot of people.

Driverless cars CAN be used when it rains (and when it snows . . .)

Paying people to drive a train is overall cheaper than implementing a 100% computer controlled system.

Some inner city metro lines in Europe are already capable of running automatically.

The technology is also probably creeping slower for trains because they aren't for sale in the same way cars are. Cars are always progressing in technology to try to get people to pay for the newest model.

Train accidents and fatalities are small in numbers compared to car accidents. The train systems are working where traveling by car could be seen as a problem.

Basically there's less of an incentive to automate trains. I'm 100% sure they can and some smaller city lines are, but there's not a lot of money in it. There's a lot of money in automated cars

It isn't near as much about the time you spend in the car, it is about what your car could do without you. You could send it to pick up your kids, groceries, etc. Uber can use it to move people around when you aren't using it.

Can_I_Get_An_Amen? wrote:

They are no where close to proving that self-driving cars are safer than human drivers. I've worked at a software company that made "sophisticated" computer vision, i.e. computers trying to identify objects in images, and guess what? No one in that company would trust their life with that software, even though it was "best in class" for the task that we created it for.

This is blatantly false. Tesla has millions (billions?) of miles of self-driving car data that do show they're "safer" than human drivers. The media will highlight the one fatal accident that happens from a self-driving car, and that's valid, but obviously they can't highlight the 10 lives it saved for that one fatal accident, since those 10 accidents didn't happen.

The ethical question isn't "Is it okay to let people use self-driving cars if they may die due to no fault of their own?".

The real ethical question that Tesla and others have to face is "At what point do we enforce self-driving because it's so much safer than human drivers?".

I think self driving cars sound fun because then I can watch a movie or something instead of driving

The best ethical questions are "Can they make them completely hack-proof? And if not, then should they ever be allowed on the road?"

The timeline for when it will be at (relatively) full adoption is hard to tell, but the bottleneck will not be on the tech but rather on the public perception and the laws and regulations (which go hand-in-hand). Fortunately the auto industry has a lot of money so will lobby pretty effectively at the city level to make sure self-driving cars are legal.

This isn't some fringe idea being pushed by kids working in a garage and eating ramen. The biggest auto-makers and the biggest tech companies in the world have been pouring significant investment into this concept for some time now, and not just in America but also in China and Europe. Europe is a secondary target though because transportation there is already, in general, much better/less of a problem than it is here.

Car ownership, and the enjoyment of it, is a generational thing, that millennials are not as interested in. Self-driving cars allows people to not own cars, but have all the convenience of owning a car.

In terms of safety, all it has to do is outperform humans, which would not be hard. In terms of traffic, tests and computer models have proven that self-driving cars offer a chance to significantly reduce back-ups caused by 1) accidents and 2) selfish people attempting to skirt as far down a backup and cut in as they can.

I for one am extremely excited about this. The biggest problem will be the workers it displaces, in the taxi/ride-sharing areas and the truck drivers. That's what, 6 million people? Possibly more. I hope the government recognizes this soon and starts directing training/assistance programs to ease the transition for these groups, just as they failed to do when companies began moving manufacturing jobs out of the country.

Xanax wrote:

...

In terms of safety, all it has to do is outperform humans, which would not be hard.

...

Nope. In terms of safety it has to be shown that these things CAN NOT be hacked. So far the idiots designing this technology have put tech security too low on the priority list and it has been shown how easy it is to remotely gain control of an autonomous vehicle.

If only a couple verifiable cases of hacking with forced crashes occur, then this technology will fail the public acceptance test, as it should.

Let me just repeat this for you idiots who continue to do this, when every bit of logic (that you obviously do not understand) would tell you that you're making everything worse for everyone else because you're a selfish prick:

2) selfish people attempting to skirt as far down a backup and cut in as they can.

DiscoGary wrote:

Nope. In terms of safety it has to be shown that these things CAN NOT be hacked. So far the idiots designing this technology have put tech security too low on the priority list and it has been shown how easy it is to remotely gain control of an autonomous vehicle.

If only a couple verifiable cases of hacking with forced crashes occur, then this technology will fail the public acceptance test, as it should.

As far as the tech is concerned, this really isn't that difficult. The current batch of automated cars are lacking in this regard, but it looks like they haven't paid too much attention to security. As the tech picks up, security will too.

DiscoGary wrote:

Xanax wrote:

...

In terms of safety, all it has to do is outperform humans, which would not be hard.

...

Nope. In terms of safety it has to be shown that these things CAN NOT be hacked. So far the idiots designing this technology have put tech security too low on the priority list and it has been shown how easy it is to remotely gain control of an autonomous vehicle.

If only a couple verifiable cases of hacking with forced crashes occur, then this technology will fail the public acceptance test, as it should.

Maybe from an irrational/emotional perspective, which I'll admit does govern a lot of public opinion and thus laws. However, if 10,000 / year get hacked and cause fatal accidents, that's still around 25,000 fewer fatalities than we're currently facing, and logic would dictate we should take the hacking risk over the more deadly one. Of course I expect hacking incidents to be much less, basically on a scale that shouldn't register.

However, airplanes have been allowed to operate almost entirely autonomously for some time now, and they are hack-able as well. There really is no such thing as "unhackable" tech. But given the stakes, and the companies/money involved, you can trust that some of the smartest engineers in the world are designing these systems, so they will be extremely difficult to exploit.

Saying the "potential for hacking" would be enough to shut it down would be like saying the "potential someone could get behind the wheel when they're blind drunk" would be enough to shut down human-controlled vehicles, and that clearly is not the case.

Why automate trains? What's the economic incentive?

There isn't one and that's why it's not being automated.

And your 60 year old dude is just plain wrong. Already Waymo is at the point where a human interaction is only required once in every 5000 miles. The frequency of interactions is dropping exponentially every year.

It's right around the corner.

If 10,000 cars were hacked per year and intentionally crashed, the technology would have to be outlawed.

Could we please just skip to the personal jet packs I was promised many years ago?

Welcome to the DG show! Where everything is made up and logic doesn't matter!

Essentially his argument is that even if it saves thousands of lives, he doesn't like it anyway so it should be outlawed. It's been astounding watching his mental gymnastics over the years.