Misfortune
teslamotors.com
teslamotors.com
It's not a fully self driving car and its obviously not strictly human controlled. Unfortunately it ends up being more mentally taxing to use this hybrid approach than to just drive yourself.
Consider highway driving, with normal human powered mode you are in full control so if you see brake lights a half kilometer up ahead you can disengage the cruise control and react on your own, everyone who has driven is comfortable with this.
With this assisted driving the car doesn't slow down right away, and its not clear if the car just can't see the tail lights lighting up yet or if its decided it doesn't need to react yet and as such you start to second guess the car,
- should I drive or should I leave it to the car?
- what if it turns when I grab the wheel? Will I make things worse by driving?
- does the car even see the object up ahead? How can I tell, its impossible to expect the car to tell you of an object that it cant' even see.
It becomes just more taxing to use the hybrid approach to driving and as such I don't use it at all. I have no doubt that the Tesla auto pilot is safer than driving, but I also have no doubt that it can royally screw up.
I've come to the conclusion that some assisted driving, like auto braking for obstacles that you will imminently hit is good but the kind of assisted driving where it can almost automatically drive for you is not really anywhere near ready, if you follow Tesla's rules on how to use it, it actually makes driving harder:(
I think you're over thinking things - just let it do its thing, but be ready to resume control if needed. I drive 90% of the time in Auto Pilot, and it makes driving exponentially more relaxing and safer.
And if you trust Autopilot to do the right thing, and get into a fatal accident, Tesla will say you should have been paying attention!
Why do we establish the experience of being driven by autopilot as something comparable to the actual action of human, hand-feet-sight-controlled driving (steering wheel-acceleration/deacceleration-dashboard/windshield/mirrors)?
We cannot trust the horse to do the same things as the car nor can we trust the car to do be same as horse.
We are on the cusp of moving responsibility from the human engineered transport (car) to the machine engineered (autopilot) transport, which transfer was previously animal to human engineered transport (horse to car).
We will believe what establishes our worldview and ignore what demolishes the same foundations.
>but be ready to resume control if needed
Let it do it's thing, relax, but be vigilant and ready to take over. It's not that hard. Watch the road, watch the dash and what it sees, if something goes awry (happens rarely) then just resume control.
Our concerns seem to appear as that driving a car does not feel like driving a horse.
Aren't we are also horses in the sense that we drive our energy in direction rather than locomotion, when we drive? We yell when the autopilot does a singular mistake where the human is allowed to make 10x or 100x without true objection or persecution.
The cage went in search of a bird. -Kafka
SEC statements are usually covering a lot of stuff, so I doubt it's very surprising.
But I'm still not convinced by their argumentation on the safety of the Autopilot.
>That contrasted against worldwide accident data, customers using Autopilot are statistically safer than those not using it at all.
I can only cite German data [0], but on the Autobahn there is a death every 90 million miles driven. But that includes drunk drivers, old cards, old drivers etc. Teslas Autopilot being save in over 100 million miles doesn't really sound that much better, especially considering the smaller sample size. And in this case, the death could've probably been avoided by not using autopilot.
[0]: https://www.adac.de/_mmm/pdf/statistik_4_5_Unfallgeschehen_S...
Some questions now are: What did Tesla do to warn of this inevitability? What did they do to mitigate it? What are they doing to mitigate it? Is there any way for users to reduce their chance of being a count in the collision statistic, other than reading and following all relevant disclaimers and warnings?
As I have noted before: https://news.ycombinator.com/reply?id=12012688&goto=item%3Fi...
TESLA really needs to tread carefully here to avoid giving the whole Self-Driving Tech a bad name.
You CANNOT move fast and Break Things in this particular case.
What does it matter if it cannot be proven to be better than regular driving?
Unfortunately, accidents with automated driving will seem a lot more prevalent if people are shown concrete cases. I believe that's why none of the players should give it a bad name in its infancy.
At any rate, it's refreshing to read rebuttals to journalists. My own experience is that it is very rare that journalists both want to get a story right and have time to get it right; the vast majority of modern 'news' is written by glorified bloggers on a deadline.
After a decade in the military, where bad things happen frequently, I couldn’t imagine public relations acting in this manner: "We bombed the wrong house, but this happens, it's war, the media just doesn't understand". Don’t give people the impression you care more about your business than their lives. Don’t dismiss someone’s death as a statistic.
I want Tesla to succeed, but they are starting to leave a bad taste in my mouth. I remember my XO saying, “never get into a shouting match with an idiot, bystanders can’t tell the difference.” Ignore the media if you can’t respond appropriately. The real misfortune here is that somebody died in what was probably a preventable accident, not a poorly sourced Fortune article.
I prefer your version, but I would like to share a slightly different formulation of that aphorism that is equally amusing: never argue with idiots, they'll bring you down to their level and then beat you with experience.
Indeed, why didn't the Tesla supplement its vision system with radar and/or laser rangefinders? It's pretty obvious that the autopilot is not ready for production use.
The fact of the matter is, a Tesla driver is not your regular driver. A better test would be to give everyone with a Honda Civic some kind of autopilot.
The data that they do have is a curiosity at best, and cannot be compared vis a vis regular joe blow accident data.
Fortune did not "assume that Tesla had complete information from the moment this accident occurred." Fortune merely noted that there was a long period of time between the time of the accident and when the information was disclosed. Fortune is not saying that Tesla did have knowledge of what happened, it's saying that Tesla should have had knowledge. And Fortune is saying that Tesla should have investigated more quickly especially considering the impending stock sale. Perhaps they did not because Tesla feels as though the death is not "material." Musk's logic on whether the death was material is so convoluted it defies any sort of logical counterargument, and is ultimately for the courts to decide.
Regarding whether or not the accident was not an Autopilot failure, of course technically it appears that the truck should not have turned, but this is what happens in normal traffic. We do not design autonomous driving systems with the idea that nobody ever breaks driving rules or laws, rather we design them with the idea that every crazy situation that we can think of, and a lot that we cannot, are likely to happen and we should attempt to avoid or lessen the impact of a crash whenever possible. When Musk says that there is "no evidence to suggest that Autopilot was not operating as designed and as described to users," I seem to have missed the marketing material where they say that the features work only when other drivers follow traffic laws. It's one thing if a construction crane drops on top of the vehicle (cannot be reasonably avoided); it's another when a truck crosses in front of you (where we can take various evasive measures to lessen the impact of the crash even if we cannot entirely stop the crash from occurring, none of which were taken in this case).
Regarding vehicle accident statistics, as someone who worked in active safety for several years (i.e. systems like radar-based crash avoidance that did not trigger in this case), when we look at these numbers to make comparisons, the entire idea as of 2016 is that when we combine a human driver with a lot of electronic help, you are far safer than driving without that assistance or if the electronics were driving for you. But Musk only compares autopilot, which is only usable in certain scenarios, with the entirely of vehicular death information (which by the way, is largely much cheaper, older cars with worse safety systems that lead to more deaths). We know for a fact that autopilot lulls drivers into becoming distracted and doing other things, taking their eyes off the road, etc, because it does not enforce drivers to keep their hands on the steering wheel (See NYTimes report today). The simple, obvious fact of the matter is that a Tesla would be far safer if it were using all of its accident avoidance technology and required the driver to be engaged with the steering wheel. Yet it does not.
As someone with experience with this technology (though not at/with Tesla), I would greatly like to see them push a software update that requires you to have your hands on the wheel while autopilot is engaged. Saying that you should do so is not enough when the tools to enforce it are there. Is is clear and obvious that that would make the technology even safer. Yet rather than accept even the tiniest shred of responsibility, Tesla has entirely closed ranks and blamed everyone else. I, for one, am disappointed in Tesla as a whole and Elon Musk in particular.
At that point, drivers shouldn't be expected to put their hands back on the wheel. They should be in the backseat, which is Google's plan.
There isn't time to put your hands back on the wheel in the event of a car accident. Planes get minutes to react to errors. Cars have seconds. Car drivers are more akin to train conductors who, these days, have complex deadman's switches that guarantee the conductor is still focused on driving the train.