Eight car pile up found to be due to Tesla driver assist software
cnn.com
cnn.com
Tesla and Musk then try to blame drivers for trusting them with their lives by saying "you must be ready to take control at a moments notice", and "you must be focused on driving" while also not doing anything - things that are challenging for professional pilots, and the latter demonstrably impossible. That (1) it is legal, (2) not grounds for a return + full refund, and (3) not grounds for criminal charges against Tesla remains absurd.
The fact that Tesla continues to get to falsely claim it improves safety, despite having been caught not including crashes where AP/FSD disengaged moments prior to crashes, and not doing the most basic like-for-like comparisons, is also absurd.
We then got to see testimony on HN today that shows that Tesla knows it is committing intentional fraud: it knows that it has aggressively marketed FSD using fraudulently created videos, and made false claims using those videos as "evidence" of safety.
The fact that they're not subject to government monitoring, and Musk is still fellated by his fanboys as a genius remains an appalling indictment of tech and safety legislation.
I thought most new cars now have automatic emergency braking and many have cruise control that can bring the car all the way to a stop if the computer thinks it's necessary. At least as of a couple years ago during the covid lull before inflation went crazy, there were new cars with this tech for barely over $20K.
So is there anything unique to Tesla involved here just because people like to talk about the "full self driving" branding? There could be, but I've failed to locate it in any of the articles about this disaster.
Plenty of other cars have emergency braking and have had it for years. But because those companies don't have a complete disregard for safety don't advertise "now with fully automatic braking!" and say you never have to brake yourself.
They also use technology that is actually reasonably robust and reliable: radar, lidar, etc to measure obstacle distances rather than ad hoc computer vision and wishful thinking.
> So is there anything unique to Tesla involved here just because people like to talk about the "full self driving" branding?
Because they aggressively push their "full self driving", and their car changed lane and hard braked for no reason and on its own, and their "self driving" tech has a history of driving people into walls, trucks, emergency vehicles, and from the article thousands of reports of random braking, now causing real harm.
Also dismissing "full self driving" as "branding", when the name explicitly states what it is, and they aggressively market it as being just that, and they produce fraudulent videos to act as evidence showing that it "works", is BS.
No other car company is as aggressively pushing/marketing anything it produces as self driving, and no other major car company has been making objectively fraudulent claims about their self driving ability, so of course when Tesla's unsafe self driving technology causes crashes people will talk about it.
What I am mildly interested in (for the next time I get a new car) is whether a Tesla (say a model 3) would be more dangerous than an ordinary car if driven by the same person in the same manner. And the reporting all seems maliciously useless. I mean "useless" as opposed to "anti-Tesla" even if those might overlap.
I remember the reporting about Tesla abandoning radar, and that's definitely a good reason to wonder how safe they are, but worrying isn't actionable.
My perception based on "anecdata" is that an ordinary car with auto braking will probably brake hard at nothing maybe once in two years. If you're lucky, not in traffic. So, I'd like to know how Tesla compares - and of course, I wouldn't trust their PR any more than you.
So if you represent your car as "full self driving", and then it causes an eight car pile-up, then it's very much the fault of the retailer or manufacturer.
Doesn't matter if you popped up a EULA with multiple disclaimers on the car's touchscreen, if you sold it as "full self driving", and it ain't, you're on the hook.
So why is Tesla not yet on the hook?
Based on anecdata, I'd expect an economy car from a boring "legacy" automaker to brake sharply at nothing about once in two years.
I haven't seen anything in the reporting on Tesla that would shed light on whether they do it more often, or in more dangerous situations.
Emergency braking is _very_ simple tech, and is actually robust.
We have a lot of evidence that they do: the article even mentions large numbers of sudden braking complaints. For other manufacturers this has resulted in recalls, and generally is due to electrical faults rather than the computer deciding it's time to stop.
We also have a good idea of why: Tesla is notorious for favoring computer vision over lidar, and their multiple fatalities involve the camera being unable to distinguish walls, barriers, trucks, emergency vehicles, from empty road. Yet Tesla is permitted to continue selling them, with that behavior and without fixing the problems. The aforementioned unexpected/sudden braking from other companies had recalls, and those recalls were required to ensure the fault could not happen again.
I didn't mean emergency braking only. I meant automatic braking to include non-emergency braking such as (but not limited to) active cruise control that can stop a car.
Both emergency braking and milder braking can and do happen on a normal non-Tesla car when the (some) computer gets confused. Electrical faults aside.
On a normal car it might happen at intervals of a year or two, so the thing I'm curious about is whether there is some evidence it's more frequent or otherwise worse on Teslas should I ever be shopping for one.
I think the quality of the hardware is only loosely related, because engineering decisions, some of which are embodied in software, could make a big difference.
If I was in charge of making hard decisions about a trade-off between phantom braking and avoiding accidents, I would say "don't make it panic brake until slightly after the computer figures a collision is unavoidable" since this still provides the most important value of reducing injuries.
But I don't know if automakers really operate this way. Maybe they know better, or maybe there are bad incentives.
And if some of them do operate this way, but others have the hubris to think they can prevent accidents and phantom braking without tradeoffs, I'd really like to be able to make an informed choice.