Tesla CEO Elon Musk has said that cars operating in Tesla’s Autopilot mode are safer than those piloted solely by human drivers, citing crash rates when the modes of driving are compared. He has pushed the carmaker to develop and deploy features programmed to maneuver the roads, arguing that the technology will usher in a safer, virtually accident-free future. While it’s impossible to say how many crashes may have been averted, the data shows clear flaws in the technology being tested in real time on America’s highways.
I dislike elon, and I’m never buying a Tesla (I own a different EV already). However, until someone shows me the equivalent human caused rates (for the same type of roads and distance) these numbers just simply don’t look out of line with what I would expect for any car.
IMO Self driving doesn’t need to be perfectly safe, it just needs to be equivalent or safer than the average human driver.
I think the bigger issue is the lack of transparency. Tesla only reported 3 fatalities involving autopilot while the real number is 17. Not a massive difference when dealing with low numbers like this, but still a big issue if Tesla is lying about safety data.
Yeah this. Maybe I could agree that it’s too soon to be testing these auto pilots on the road, but I dislike how people miss the point with this tech. They set an impossible standard for a technology that could potentially be better than us on the road
@BlameThePeacock @tango_octogono Fair, and in principle I agree. But it’s not (only) the people who miss the point and set an impossible standard but foremost Mr Elon Musk himself. He has been promoting Tesla’s autopilot and even its self-driving capability for years (although the folks at Tesla will certainly know that the latter won’t come anytime soon).
Tesla video promoting self-driving was staged, engineer testifies
A 2016 video that Tesla (TSLA.O) used to promote its self-driving technology was staged to show capabilities like stopping at a red light and accelerating at a green light that the system did not have, according to testimony by a senior engineer.
One of the things we needed for setting reasonable expectations regarding this tech is more reliable information also from Tesla and its CEO. As long as the company itself is frequently flooding the market with unrealistic “news” about this tech, it is good that there are independent investigations imo.
They set an impossible standard for a technology
Maybe that’s the case but it’s not what the article says. Experts say that “the surge in Tesla crashes is troubling” and that “the number of fatalities compared to overall crashes was also a concern”. And they are critical of Tesla as the company is obviously beta-testing a car on the highway without disclosing its data as others have already said.
I don’t see someone setting impossible security standards, at least that’s how I read the article. Tesla appears to value its profits more than the safety and lives of people.
I was talking in general and not necessarily about this article alone, although it plays some notes that are common in stories/articles critical of this technology, like using personal individual stories to appeal to emotion. It’s an article so yeah I expect them to do that, but when it comes to safety it’s better to look at the numbers, as every option will have tragic stories.
We humans suck at picking the better options if we let subjectivism and emotions take too much over the discourse. The US for example had a “war” over seat belts for example.
As for Tesla, not at all surprised about the points the article rose though. Few weeks ago we found out that Tesla asked employees to not register complaints on paper, and there was also the fiasco where employees where sharing around private images of their customers. Tesla is a sick company and I wouldn’t be surprised if they have more skeletons in their closet.
Please read this thread to understand the issue: https://mastodon.social/@adamjcook/110505648591695279
I agree with you that we need more data. Right now the USA average car death’s are 1.37 per 100 million miles driven.
From what little Tesla has talked about the autopilot is below that average. BUT the raw data hasn’t been released. We don’t know how many miles have been driven on autopilot, we don’t know the road conditions it was used in (assumingely autopilot would be used more often on freeways), and we don’t know how the safety rating of the Tesla vehicle compares to others on the road (its possible Teslas are getting in more accidents but the car is keeping them safe, or vice versa).
Too many unknowns. So while I dislike this article because it mostly comes off as hollow in my eyes, I do think that Tesla needs to make more of it’s data public so users can make an informed choice.
Yeah, the article reads like a hit piece. Why didn’t they even try to do an apples to apples comparison? Same thing when they compared Tesla to Subaru, just raw total numbers. Not normalized to cars sold or miles driven or anything. The raw data is pretty meaningless.
Why didn’t they even try to do an apples to apples comparison?
Maybe because the data released by Tesla is incomplete and biased as it appears to serve its sales rather than safety?
It’s the company and Elon Musk himself that are frequently making bold statements while it seems that not even the authorities have the data to verify the claims. As the article says:
In a March presentation, Tesla claimed Full Self-Driving crashes at a rate at least five times lower than vehicles in normal driving, in a comparison of miles driven per collision. That claim, and Musk’s characterization of Autopilot as “unequivocally safer,” is impossible to test without access to the detailed data that Tesla possesses.
I’m quite sure you’re right and the data Tesla releases is biased. Which makes it all the more important to make unbiased analyses. Unfortunately raw accident numbers that aren’t normalized to anything are useless.
Yeah, this article is ridiculous. Self-driving cars are FAR safer than human drivers. The number of accidents is minuscule compared to what would be expected if the automated features were absent.
We run the real risk of screwing this up if people insist on an automated car never causing a crash, or never hurting anyone. Even if they hurt someone, the point is that it harmed maybe 0.1% of the people that would have been hurt in traditional vehicles.
”Don’t let perfect be the enemy of good”
Self-driving cars are FAR safer than human drivers. The number of accidents is minuscule compared to what would be expected if the automated features were absent.
We run the real risk of screwing this up if people insist on an automated car never causing a crash, or never hurting anyone. Even if they hurt someone, the point is that it harmed maybe 0.1% of the people that would have been hurt in traditional vehicles.
i feel like you’re illustrating the issue here but from the other direction. are they? is the number of accidents miniscule? self-driving technology is frequently hyped up in exactly this manner–particularly by Elon and Tesla apologetics, who have a vested interest in it being correct–but i’ve seen nothing to suggest that the technology is either widespread enough or reliable enough to draw a meaningful conclusion in either direction.[1] i also don’t think it’s reasonable to conclude, if there’s an absence of numbers, that these kinds of technologies are just inherently more safe than humans. we’ve already seen plenty of technological snafus that have potential to be way more harmful at scale than anything the human brain can muster.
as far as reliability: Tesla self-driving technology struggles in a lot of cases, and even defenders of the cars will admit it has problems in many circumstances interpreting basic rules of the road and pedestrians in its path. this is obviously a problem now, and would be a much bigger one at scale. ↩︎
Tesla apologetics, who have a vested interest in it being correct
I think this is a little dangerous way of arguing. Now suddenly everyone that argues in favor of Tesla is an apologetic with a vested interest in being correct?
Everyone making a forum post have a vested interest in being correct, and you can make arguments for a case without being an “apologetic” / apostle / evangelist / fanboy etc.
In short, I don’t like this way of making arguments and trying to group all your argument opponents into some non thinking hive mind.
In short, I don’t like this way of making arguments and trying to group all your argument opponents into some non thinking hive mind.
i mean, to be very blunt, if the Tesla apologetics want to not be grouped in that way they could start by not acting like a non-thinking hivemind. the issues with Elon and his products are well-stated, well-founded, well-established, and indisputable. Elon himself is a categorically awful person who wants to make the world a worse place and frequently undermines actually useful projects with techno-fetishistic, autofellating “solutions” that demonstrably don’t work. he has maybe one project you can actually credit to him on any meaningful level (SpaceX) and on a good day his impulsiveness and reprehensible politics still get in the way of the good work people at that company do. literally anyone else would be a better leader for the company purely on the grounds that they would not be a distraction from that work. he was born rich and has parlayed a series of failures into being the richest, dumbest man alive–by any objective standard he is the classic person who has failed up.
but none of these critiques mean anything to his core fanbase, because they don’t care and are fully bought into his mostly-manufactured cult of personality. there is no point in trying to argue this with that group; you can cite any number of the things i just listed at them and it rolls off them like water. frankly a lot of the time in my experience they’ll do the cryptocurrency guy thing of accusing you of sowing FUD, or being jealous of Elon, or just being a woke lib because you want public transportation and not the masturbatory self-indulgence that is the “Hyperloop”.
Okay but none of that is relevant to the question of “Are computers in their current condition better drivers than humans?”. No one here thinks that Elon is cool or good, I haven’t seen a single fanboy comment so far. But that doesn’t change the hard facts about the things that a couple of his companies (who have hired a lot of very good engineers) have accomplished. Pretending that Tesla and SpaceX have zero accomplishments or benefits does not enhance your arguments against their boss, it detracts from them by discrediting you. There are plenty of legitimate criticisms to make, many of them absolutely damning, there’s no need to falsify anything to condemn them.
Let me just keep it civil and say that we absolutely never will see eye to eye, and I have no interest in trying to change your mind. I’ve said my peace on your argument technique and definitely stand by it after your reply, and I’m happy to leave it there.
@freshhotbiscuits
Even if they hurt someone, the point is that it harmed maybe 0.1% of the people that would have been hurt in traditional vehicles.
Where have you got this number from?
This is exactly what I meant in my comment above. Anyone throws out a number and claims it is true. I don’t think we should let perfect be the enemy of good, but here are people’s lives on the line. We need independent and reliable data.
Oh, I wasn’t stating that as fact, or ever intending to. The fact of the matter though is that more than 90% of car accidents are caused by human error. If we can eliminate the human error, then we’ll have far fewer people hurt on the roads, even though that means that self-driving cars are going to hurt people. This is merely my point, that we can’t expect self-driving cars to have 0 accidents, but once the tech is good enough (and I acknowledge that it’s not even close yet), we need to be ok with that hard truth.
but once the tech is good enough
What makes you think that the tech is already good enough for being tested in the public? The fact is that the incidents revealed by the investigation is much higher than what Tesla has published, Tesla is holding back a lot of relevant data, no one (except Tesla) can say how safe this tech is and whether or not it should be allowed to be tested on the streets.
The fact that many or most car accidents are caused by human error and a lot of other critical points here simply don’t matter here as they have nothing to do with the issue. This is not some application on your smartphone that you can test at your own risk while it is still in beta. This is a car. It kills people, and Tesla is obviously unwilling to disclose the data even to the authorities. As long as this is the case, this tech should not be allowed to be tested in the public space.
