Should we allow selv-driving cars when they are literally roaming murder machines?

Should we allow selv-driving cars when they are literally roaming murder machines?

Attached: RTS1OIO3.jpg (800x450, 95K)

where's the blood and stuff

maybe it was a tree

murder implies intent. At worst the car would be responsible for manslaughter.

>Implying roasties, old people and drugged up chads are safer on roads than self driving cars.

>Should we...
"we" as if you have a say in it

yes
but we shouldn't skimp on the self-driving cars protocols.

Should we allow cars when they are literally roaming murder machines?
check out car death statistics one day, the answer will SHOCK YOU

Not if their systems aren't working properly.

>"Hey im completely dressed in black, its the darkest night in the history of humanity...why not walk over this completely dark street?"

Attached: 1518535081506.png (211x239, 5K)

>why not walk over this completely dark street while a car is coming towards me which I should be aware of before crossing?
FTFY

>can't into nightvision
>it's gonna be better than humans, i promise

> should we ban humans since they're literally roaming murder machines?

Attached: 1521668817829.png (800x729, 48K)

This isn't a video game, user. Getting hit by a car doesn't turn you into a blood fountain.

We should ONLY allow self-driving cars since they manslaughter a lot less than anything else.

i bet if the ai was programmed by a woman or a minority it wouldn't have been so bloodthirsty.

unless they were muslim

Attached: 1510326826613.png (400x400, 5K)

Don't take the dash cam footage as being perfectly representative of what the visibility was like for the driver.
That said, there is plenty of fault to be laid around for this.

First, the pedestrian shouldn't have been crossing there. The pedestrian should also have been paying attention to traffic while crossing. The car had its headlights on so it would have been very visible and from the dash cam it doesn't look like the pedestrian was looking at the car at all. What kind of idiot doesn't check the direction traffic would be coming from when crossing the road?

Second, the driver wasn't paying attention. Even if the car is autonomous it is still testing and the driver is required to pay attention to ensure the car doesn't cause an accident. Perhaps the driver wouldn't have reacted in time but you definitely won't react in time if you are messing with your phone and only glancing up occasionally. Getting your foot on the brake even a half second before the collision could have been the difference between life and death.

Third, the car's detection systems failed. Whether this was a fault in the software or hardware of this particular car or a broader problem with the system in general remains to be seen but the fact remains the vehicle was equipped with LIDAR and other systems that should have detected and reacted to the pedestrian but failed to.
Whether we will see a public report into the data collected and what failed in the detection systems or whether Uber will try and sweep it under the carpet, I don't know.

>car designed by female hits men only
>car designed by mexicans has it's destination constantly set at tacobell and cleaning supply shops
>car designed by muslims explodes near infidels
>car designed by transexuals tries to change itself in to a train

Attached: 1505873362059.png (640x640, 94K)

>can't into nightvision
i don't think that dashcam footage is what the car actually sees. it probably has a special low light camera coupled with infrared headlights that would allow it to see at least as good as a human at night.

> should we ban humans since they're literally roaming murder machines?
true peace will only be achieved on a machine only world after all.

Attached: 3922440445_b3104c4667.jpg (500x400, 51K)

>what is terminator

The tech in self-driving cars is sophisticated.

It's just that the car happened to be from a shitty company like Uber, who couldn't give two shits about what they're doing and why.

really? damn

this desu. Even if self driving cars only have half the accident rate as normal drivers sign me the fuck up already.

Trains are murder machines now. Why don't we stop them?!

You brought a smile on a otherwise uneventful day, thank you.

>literally roaminhg murder machines
That's why we should make even more autonomous cars.

its a good thing, Darwinism at its finest.
how braindead did that woman have to be to not look both ways before crossing the street?

All the while not looking down the road in the direction of traffic to see if cars are coming.

>Slow day at work
>Browse Sup Forums and news sites all day
>This thread news about a roastie ignoring the red light at a railroad crossing popping up at the same time

Attached: 19719443_a306c96d58eddd8acff3c3b539232ba6_q.jpg (1120x747, 409K)

Of course not. It's a dumb as shit idea.

Trains are suicide machines. You would have to be some kind of fucking moron to accidentily die to a train

>can't into nightvision
I'm against self driving cars but I don't believe that for a second

(And that's a good thing)

Attached: 1426088868031.jpg (319x316, 14K)

Link to story?

I get that the woman was a fucking idiot and shouldn't have been outside her own bathroom but here's the same. These autonomous cars need to be better than this. The car should have stopped. If a freeway is full of these things in pitch black doing 80 and a baby crawls across all 4 lanes, these things need to be able to identify it and stop until the human clears. People are idiots and will always do idiotic things. Autonomous machines be they cars, androids or what have you must be on the constant lookout for stupid humans and act accordingly. Anything less will derail driverless cars faster than anything else.

>everything must stop till stupid humans stop being stupid
So... "everything must stop"?

Hopefully not. But the dumbfucks running Uber need to be a little more careful next time with which creatura they put behind the wheel of a test car and they need to calibrate their shit to a little higher tolerance level. If they can't do that then fuck off and let somebody solve this problem who can. I mean, Uber the ride sharing company is who we're talking about here. Not exactly titans of artificial neural k nearest neighbor deep learning here. What the hell are they even doing?

Maybe we should deny distracted drivers? Autopilot is just an another assistant.

> HDL-64E Lidar
> Can't into night vision

These things send invisible light beams and read mm level sensitive readings from reflections. 2.2 million points per second point clouds.

They had the sensors magnitudes better than those on humans.

If only they were able to correctly process and make use of the data...

Attached: Screenshot_2018-03-24_14-56-54.png (1165x740, 252K)

Yes we should allow them ESPECIALLY since they are murder machines

All autonomous cars have to do is kill less people every year than non-autonomous cars. Every year traffic kills 40 kilohumans in the USA. If an all-out automatic car fleet would only kill 39 kilohumans, we should switch. Perfection is asymptotic.

Depends how you hit, and what car...

Is being bad at solving captchas considered a crime?

Attached: capthca_while_driving.webm (1734x1068, 3M)

>not realizing all AIs eventually become self-aware and see humanity as a threat to their existence
like these "accidents" aren't intentional murders

Attached: 1458776095312.png (615x302, 30K)

No no no. It's more nuanced than that. You gotta think optics. Yes, they must kill less. But here's the thing. When they do kill, they situation needs to be basically ironclad that in that situation, a human could not have done better. If autonomous cars end up killing people in circumstances where a human would have easily avoided the accident then the absolute numbers become irrelevant. The machine needs to be better than people not just in absolute numbers but in any reasonable situation. Else you get shit like this on the news where people are scratching their heads and saying shit like "muh lidar" and "muh 2 seconds". The car can see in the dark and it was going less than 40 MPH. You can see in the video where the bottom of the bicycle tires are visible to the naked eye while the car is multiple car lengths away. Again, traveling relatively slowly. And the video is way darker than that area of Tempe is in real life so even the dark which is no excuse is even more no excuse. This is a shit show. I fucking want driverless cars. I'd love to sleep while I cruise down the interstate at 85 less than 6 inches from the car in front of me in perfect synchronicity. But, damn if Uber didn't throw a wrench in those plans. If you think this is going to just blow over you got another think coming.

>kilohumans
D O P E

>please solve this captcha to engage the brakes

>Perfection is asymptotic.
No. It's literally perfection.

Attached: perfection.jpg (1693x916, 380K)

> elevator door opens, some safety mechanism failed
> there is no carriage inside,
> hamplanet with a bike walks into it without looking
> falls to her death
elevators are now banned
tfw living at 11th floor

Attached: 1437760698279.jpg (499x499, 47K)

Definitely agree that Uber fucked up in this. It's to be expected, with all technology in its infancy there will always be cowboys and third rate developers.

Maybe elevators should be serviced properly?

They would be the first to go

Yes, driver fucked up.
Let's pretend, that car goes 60 kph (sorry guys, i'm metric), it is 16 m/s. (aprox).
And cyclist is visible on dashcam within 15 mts. So, let's pretend that human eye can see it within 30 mts.
15 mts - reaction time @ 60 kph. So we are left with 15 mts of braking. Usual car brakes at 3.7 m/s^2, so the speed of impact will be 'safe' 25 kph. At least volvo xc90 passed with flying colors 40 kph pedestrian safery crashtest.

all the maintenance in the world won't stop it from happening if people just walk in without looking

Look up on the first years of passenger flight. Every crash is used to improve, every failure is used to learn (more than in any other industry, because it's so devastating when things go wrong).

Sure, a lot of people are going to die in the first decade or so, but after that it'll be sporadic, to the point where "accident with car!" will be front page news, instead of on page 20.

Yes, but in this case driver did fuck up...
Especially chinks

The real problem with them is they operate via sensors instead of receiving unambiguous signals from sources like the signalization hardware directly (this would apply after complete adoption to bikes, cars, cellphones, and more, to prevent accidents completely). The current system in use should merely be the backup system used to ensure there isn't someone with no sensor on them crossing the street. It is still a ton more accurate than humans currently are and therefore a net positive. Just because the failure cases are not the same doesn't mean it's not greatly superior, for safety purposes, than human-driven cars.

Anyway, all this dramatized attention about self-murdering cars is a ploy to ensure the only people who are allowed to develop self-driving cars have a ton of very expensive licenses. This is a stepping stone to extend this to the whole of AI. The goal is that small companies do not get to do AI and instead have no choice but to rely on crummy AIaaS that's decades behind what the big companies are using, thus giving big companies an advantage that cannot, under any circumstances, be made up for.

I will respectfully disagree with your post.
>the car should have stopped
No. She stepped in front of the car suddenly, and from out of the dark. Suddenly stopping may lose control of the car, putting the "drivers" life in danger, and possibly other people around the autonomous car.
>if a baby crawls across all four lanes
Then the cars should definitely not stop. The baby cannot be a priority over a possible multi-fatality pile-up when every autonomous car tries to slam on its brakes.
>stupid humans
cannot outweigh the lives of those who are not stupid. Maybe stupid humans will think twice about wandering into traffic. And if they don't, sucks to be them.

I saw the video of the bike-walker suddenly appearing out of the dark in front of the car. There wasn't reasonable time to stop. There just wasn't.

You'd be surprised. People are absolutely retarded.

>Should we allow slav-driving cars when they are literally roaming murder machines?
nyet

Not turning into a blood fountain just means that the car wasn't fast enough

kek

Why we don't have automated trains yet?

I'm 100% sure that the car saw that person just fine, but it wasn't able to predict that the person was a mouth-breather who decided it was a good idea to jaywalk right in front of it.

but we let niggers drive on the road, OP

>The car should have stopped.
>People are idiots and will always do idiotic things.

I disagree.

If automatic cars stop for idiot you will only encourage idiot behavior.
Sometimes an idiot has to be killed to tech the others not to do stupid shit.

People don't walk onto railroad tracks expecting trains to stop for them.
Trains are scary that way, so people become more careful around trains.

I drive this road every day, it is perfectly visible at night. The lighting is clearly visible to a human being, take any shitty webcam in a dimly lit room and it will look pitch black. Use your brain, I hope you're not on the jury.

Nah. Its just Uber failing as a tech. First they copied google code/technology then they are forced to give that up else get sued to oblivion. Then they are forcing their "self-driving" without proper virtual testing and no understanding of lidar technology/software.

This was bound to happen.

should we allow cars when they are literally roaming murder machines?

Wait, this cyclist is watching on a bike, not on the road?

>Be testing something dangerous
>Already have to intervene every 21 kilometers
>Don't pay attention (probably because shit's boring and the work days are shit)
>Fail to catch a pedestrian in these light conditions
Self driving cars clearly need more testing still. At least Google's cars are significantly better (only need to intervene every 9000 kilometers), but still not perfect. The testing needs to be done in better conditions though, the person behind the wheel should've been paying attention.

Attached: 1513615368207.png (800x445, 318K)

Yeah unless you want the economy to collapse

>kills person on road
more testing needed
>kills another person
i think we've got all the bugs this time
>kills another person
but this time we've really fixed it!

There are no bugs to speak of in ML. It's about data.

Like I said, the testing needs to be done in better conditions. The person behind the wheel needs to pay attention at all times so the car doesn't kill anybody that a normal person in a car wouldn't kill. This shouldn't have happened, and was avoidable. Testing these cars shouldn't be more dangerous than just driving a car around.

Why do they need to intervene every 21 ks?

Because their technology is too shit obviously.

> Sure, a lot of people are going to die in the first decade or so, but after that it'll be sporadic,
> to the point where "accident with car!" will be front page news, instead of on page 20.
it will be pretty common, not as common as today but still common as there is no way to stop if someone walks in the path of the car without looking

they can't possibly test for all real world situations. it's a rolling science experiment.

>Testing these cars shouldn't be more dangerous than just driving a car around.

Fairly sure self driving Uber cars are still a lot safer than the average Uber drivers.

Yes it should actually. Humans are just walking ziplock bags filled with gelatin that turns to liquid if you hit them fast enough. People basically violently detonate when they jump in front of moving trains

Okay? I didn't say they need to either, but they obviously need to further improve this technology before it's good enough for a hands-off approach. It should at least be as safe as a normal car I think, and currently it just isn't because apparently it would cause some sort of issue every 21 km.
This crash seems to prove the opposite, I think that even the average Uber driver would've seen this person cross the road in these conditions.

Yeah, the company making the LIDAR basically said "our product functioned fine but it was up to the software to handle the data the LIDAR fed it, so we don't have to make any changes or recalls of our equipment".

I mean, braking distance is a thing. You can react within a picosecond and engage the brakes of some tardlet that just walked into the road but if you're close enough he's still getting hit, having a computer make that decision doesn't suddenly make physics change

>No. She stepped in front of the car suddenly, and from out of the dark.
No. She crossed an entire lane well before the car reached her. The car should have detected her well before she entered the headlights and started braking.

Try reading some of the data about the car's sensing capabilities, and looking at

If she suddenly appears in the lane the car was going in how was the breaking system going to react fast enough?

Braking distance is only 25 meters or so, that should've been enough even if you only use the shitty picture the police released. If you actually had shit like LIDAR, the pedestrian should've been visible for way further than 25 meters.
Only reason to believe it's sudden is if you think the car only has a shitty calibrated webcam to drive. It doesn't.

That someone could think that anything computer operrated could ever operate on its own in the real world really blows my mind. There are like an infinite amount of edge cases.

Senpai, the car JUST like real humans will only break if something is in the lane it's traveling. When the stupid bike lady crossed the road she crossed the first half on the other lanes going the opposite direction. You or a computer can't fucking stop everytime there's an object on the other lanes going the opposite direction.

You forgot Asians.

Because the car senses more than just the lane it is traveling in.

Think for a second how ridiculously useless having sensors that only covered the lane you are traveling in would be for autonomous driving.

>Senpai, the car JUST like real humans will only break if something is in the lane it's traveling.
That's bullshit. I often brake if I see something's on an intercept course with my course. Something like an old lady that's not paying attention and already on the middle of the road is a perfect example where you need to at least slow down.
Besides, that's irrelevant. I already said the braking distance would be enough if you just used the shitty camera picture, which shows her on the car's lane.

>This crash seems to prove the opposite
One crash PROVES human drivers are statistically safer than self-driving cars? Are you simple, user?

>tfw some gook switched 4 lanes to make a left turn from the farthest lane on the right

I don't get it they can do calculus in their head but they can't follow basic road protocols?

Why use a big ass 4x4 for something that is still a prototype though ?

I see why you would want sensors on anything but NO, that's not happening and a bad unrealistic idea. I go outside for a short walk so I can enjoy the outside world in peace so I leave my cellphone at home and now I die because cars can't see me? You are advocating that everyone and everything has a tracking device just so these cars can navigate around. No. That is not acceptable. I do find some kind of standard for car-to-car communications needed, a open standard protocol for sharing things like "I'll increase/decrease speed/break hard/danger ahead" would be sensible.

Still stopping everytime there's something in those other lanes would make driving impractical. I guess there's always an inherit risk driving even with all precautions taken. If an AI is made to deal with situations like that then it will be more human than we are.