"This incident was due to human error. This is why we believe so much in making the roads safer by building self-driving Ubers. This vehicle was not part of the pilot and was not carrying customers. The driver involved has been suspended while we continue to investigate."
https://techcrunch.com/2016/12/14/uber-looking-into-incident...
Also: Uber is not listed in the CA DMV list of companies allowed to test autonomous vehicles? https://www.dmv.ca.gov/portal/dmv/detail/vr/autonomous/testi...
https://newsroom.uber.com/san-francisco-your-self-driving-ub...
Much effort is expended on branding and naming. This would, at a bare minimim, include a basic consideration of how a word is commonly perceived, irrespective of whether the common perception is accurate.
Tesla is fully aware of the cachet that the name confers upon the brand, while displaying it prominently and relegating the absolution and disclaimers to the finer print. I like Tesla. I admire Musk's acumen and vision. But, here, they are certainly equivocating and it's irresponsible.
You are being disingenuous at best.
Uber Legal: "assisted driving"
Pick one.
My new side project is a The Aristrocrats joke, except it ends with "The Gig Economy!"
And, of course, it begs the question: what would their response have been if it was due to a technical glitch, which may yet happen? Their statement could easily be used against them.
That's not a problem at all. They'll just use "git blame" and fire the programmer whose "human error" caused the accident.
If thats true I wonder why the cop didn't pull the uber over.
edit: This also leads to another question - how does a cop pull a self driving car over? Does the car know to respond to police lights? And if it does pull over then who does the cop talk to?
https://techcrunch.com/2016/12/14/uber-looking-into-incident...
The car certainly should know how to respond to police lights/sirens. I was reading Google's monthly reports on their cars and they were talking about adding a pull over response to their system.
It's going to be even funnier when the patrol car is an unmanned self-driving bot itself.
Sure, every user has some threshold beyond which they would not be fine with it anymore, but the bad boy image Uber so carefully crafted over the years provides a considerable amount of self selection so that those who would care most never even install the app.
I almost always give 5 stars, but with the notable exception being distracted or aggressive drivers.
Sorry for ranting, I wish more people would give realistic ratings to things instead of starting with 100% perfect and decrementing for misses. It should start in the middle, 50% and go up or down based on experience. This would give a much better set of ratings that mean something.
No, it's a statement that the quality is what the customer would rate as a 4 on a 5 point scale. If Uber'a expectations of ratings are unreasonable based on the way it's customer base assigns ratings, then it is Uber, not customers, who should adjust.
Its the same problem with the game industry has with reviews. If I were to market a game as "so and so" gave it 7/10 no one would care but if I say 9/10, 9.9/10 or better yet 10/10 than we have a marketing headline.
I wish more people would understand this.
problems now: we don't have enough flexibility to give 4 stars for a good but not perfect ride
If we didn't have this rating problem, 4.5 threshold would probably be much closer to the 50%.
Similar problem many professors face in schools with grade inflation. You are not holding any system accountable by using your own sensible scale instead of the one that matters, you're just punishing the student.
If you think of the service as a means of conveniently getting to a given destination and not as a performance in artfully begging for a good rating, then 100% should not be too unlikely.
Once you start giving extra points for nonessential extras, a driver who is polite and always on time will get a lower average rating than one who is 20% unreliable but hands out warm towels and free cake.
The Uber driver who just moved to the city and doesn't know his way around without Google maps is going to have a difficult time in major metropolitans.
In fact it would be interesting to see how ride times compare between driven and driver-less Ubers. I suspect that initial results will show that driverless cars are slower and that this might make some people prefer human drivers.
One of the main functions of that human is to accept blame (or at least obfuscate culpability) and absolve the tech in the event of an accident.
There are a ton of scenarios wherein it would be virtually impossible for a human--especially one lulled into inaction and passivity for most of the ride--would be able to intervene quickly enough to avert an accident once it became apparent that the tech was failing.
Humans who are lulled into passivity by highway driving don't get to escape the blame if they crash because they couldn't respond quickly enough if the situation changes suddenly, regardless of 'tech failing'.
If you have SatNav and it says 'drive down here' and you do, and it's a one-way street, you don't say "that's exactly what Garmin want, a human to absolve them when their tech fails". Because the driver is in the position of responsibility, the tech is not.
Because the "implied-Evil corporation" offers technology which claims to be capable of doing something, entreats you to confer upon it some degree of trust, then holds you responsible when it fails. Because lives can be lost as a result of such failures and the company's stance represents a cynical and cavalier regard for this fact. Because, they want the cachet (and the cash) without the responsibility.
>Humans who are lulled into passivity by highway driving
That analogy isn't even close to holding. In that case, you are understood to be in control of the vehicle. In fact, you are actively giving minor corrective inputs at all times. It's nowhere near the same as the car saying, "here, let me do that for you", but, you are supposed to sit at the ready, hands hovering slightly over the wheel and foot over the brake, ever scanning and ready to take over? That's asinine, and it's a flawed model. At a minimum, your reaction time will be slowed as you must first recognize that the car is not responding properly in, perhaps, a split second when it suddenly does something unexpected.
Look, it's simple: the car needs to either be fully ready and safe, or it's not. This have-it-both-ways Tesla-invented PR spin that says "our tech is statistically safer, but it's your fault for not correcting it even if it does screw up" is a pernicious subterfuge and I personally don't like how it defines our relationship with technology.
IMHO, every self-driving car must be equipped with a black-box to capture all information in case of an accident and special commission for investigation, to make rules to follow, so self-driven cars will be safer for others, and that is all. Self-driven system manufacturer can be punished only if these rules are not followed, by suspending a license to manufacture self-driving systems. Currently, we have almost no such rules at all.
IMHO, the first rule for self-driving cars, until they will be mature, must be better communication between computer and human: as driver, I want to see how computer sees road and obstacles, and planned actions, and other variants of these actions to pickup. I want computer to project computer map to wind-shield and be presented with tree buttons: big red «stop» button, left button, and right button, so I will be able to see road situation, computer understanding of that situation, and will be able to correct computer or make a full stop.
Are you suggesting that the 'something' it claims to be capable of doing is, explicitly, "driving 100% without human intervention"?
because you seem to be pushing "it claims to be able to drive without human intervention and cannot", which would be misleading and probably grounds for lawsuits about fitness for purpose.
but ... I don't think that's what it claims to be able to do.
In that case, you are understood to be in control of the vehicle. In fact, you are actively giving minor corrective inputs at all times. It's nowhere near the same as the car saying, "here, let me do that for you", but, you are supposed to sit at the ready, hands hovering slightly over the wheel and foot over the brake, ever scanning and ready to take over? That's asinine, and it's a flawed model.
You, as the human driver, are responsible for control of the vehicle. If you find the terms of some assistive device 'assinine' because you won't be able to be in control, then do not figuratively click 'I agree to the terms and conditions'.
You, the responsible driver, would never believe a child telling you they were capable of driving safely, it would be dumb, you know it's not true - and you're not legally allowed to hand over responsibility for vehicle control to them anyway.
Like you're not allowed to hand over responsibility to a piece of software, no matter how enticing or convincing the marketing blurb seems. And if you do hand over control to it, and then fail to take back control and an accident happens, you should be responsible.
Look, it's simple: the car needs to either be fully ready and safe, or it's not.
Look, it's black and white thinking which doesn't reflect reality. But if you want to force it to binary then it always casts false. The car is never 100% ready, you never hand over control, problem avoided, case closed.
Saying "it has to be 100% ready, I don't believe it's 100% ready, but I'm still going to hand over control, even though I think what I'm doing is asinine because I can't respond quickly enough - and then it crashes which is totally not my fault because I knew I couldn't take over in time!" is have-cake-and-eat-it nonsense logic.
That's exactly what it's attempting to do at a given point in time. It's just disclaiming fitness and holding the human driver responsible for ensuring that it doesn't screw up.
>find the terms of some assistive device
You've been had. You're buying into the alternative terminology scheme that oscillates between disclaimer words like "assistive" and sexy PR words like "Autopilot". And, the reality is that the car is not "assisting". It is taking full control from the driver and completely operating the vehicle at significant intervals; leaving the driver only to monitor it. I don't know how to make it any simpler.
Beyond that, it's no secret that these companies are trying to achieve full autonomy, and they are already making claims that they are safer than humans, even denigrating human drivers in the process. But, they want those same fallible human drivers to be responsible for taking over when their tech fails. That's what's asinine.
>You, the responsible driver, would never believe a child
>you're not allowed to hand over responsibility to a piece of software
Huh? I'm not sure if you're completely missing the point on purpose, but there's really not much more I can add.
https://techcrunch.com/2016/12/14/uber-looking-into-incident...
The point is that we don't know that. It may have been under driver control at the time.
There's probably a good argument for having some sort of external indicator of whether or not a self-driving car is under computer control or human control. Maybe a small roof mounted beacon that has to be lit up or something.
https://techcrunch.com/2016/12/14/uber-looking-into-incident...
We don't want to end up in the situation where car manufacturers make quick updates to cover flaws.
If the software is broken, the cars should be grounded, the software should be fixed and tested for at least X hours of driving.
https://www.youtube.com/watch?v=_CdJ4oae8f4