Ralph Nader urges regulators to recall Tesla’s ‘manslaughtering’ FSD vehicles
theverge.com
theverge.com
FSD is dangerously immature software that is not anywhere close to road ready, and yet we are supposed to trust it not to kill or injure people.
Some examples:
- Teslas can't distinguish yellow stoplights from the moon.
https://twitter.com/JordanTeslaTech/status/14184133078625853...
- Numerous examples of near-misses where it decides to veer into pedestrians
- They are demonstrably much more dangerous than other cars:
https://www.tesladeaths.com/resources.html
Aside from the software, they also aren't safe on a mechanical level either, since for instance, they don't have good mechanical safety mechanisms as a fallback for when the electronics fail, and people are trapped in burning cars.
https://www.autoblog.com/2019/02/25/fatal-tesla-model-s-cras...
In my state FSD/Autopilot luckily isn't street legal, but if it was and I saw another Tesla on the road, rather than take my chances I'd pull over if I wasn't able to keep sufficient distance from them. And I certainly wouldn't want to take my chances near them as a pedestrian.
I will give Tesla credit for popularizing EVs and innovating auto design, but that doesn't make up for such poor execution in manufacturing, safety design, and alpha-quality software being trusted to make decisions that can kill people when.
It's complete willful ignorance to participate in this well-funded paid fearmongering campaign driven by a Musk opponent who ran for office just so he couldn't be sued for libel and defamation for his statements.
https://www.cbsnews.com/sanfrancisco/news/dan-odowd-senate-c...
The paid anti-Tesla ad I just saw on YouTube is also an advertisement for the guy's own software company. Self-driving cars are the future and are already saving lives. (https://www.teslarati.com/tesla-autopilot-eight-times-less-l...) Promoting this Luddite anti-self-driving fearmongering campaign will directly lead to more deaths.
I don't have teeth on either party, but I want to see independent results instead of self-reported results. It seems that Tesla isn't that willing to allow independent analysis though.
https://www.motortrend.com/news/nhtsa-tesla-autopilot-invest...
Check out NHTSA's safety ratings at https://www.nhtsa.gov/ratings.
In fact, NHTSA has multiple investigations open into FSD and Autopilot because of recent fatal crashes.
Plus, NHTSA was responsible for the misleading claim that Autopilot provided a 40% increase in safety after the Joshua Brown fatality, but that number was later torn apart by independent researchers after they sued for the data behind the claim, and NHTSA backed down.
https://arstechnica.com/cars/2019/02/in-2017-the-feds-said-t... https://www.reuters.com/article/us-tesla-autopilot-idUSKBN1I...
In the 4th quarter, we recorded one crash for every 4.31 million miles driven in which drivers were using Autopilot technology (Autosteer and active safety features). For drivers who were not using Autopilot technology (no Autosteer and active safety features), we recorded one crash for every 1.59 million miles driven. By comparison, NHTSA’s most recent data shows that in the United States there is an automobile crash every 484,000 miles.
Unfortunately, it doesn't look like they split out numbers for FSD, which is important to understand in this context.
Assuming FSD is currently as good as a human driver (Tesla suggests it is better), how much better does FSD have to be than a person to be accepted as a safer way to drive?
That's a massive assumption made without evidence.
> Tesla publishes details about accidents per million miles driven and compares that with Autopilot active and without.
The problems with Tesla's autopilot (different from FSD) statistics have been pointed out before: https://arstechnica.com/cars/2022/06/teslas-using-autopilot-...
I find it of an accusation that you say I've a "massive assumption made without evidence". The "evidence" I have is the data from Tesla itself, which states that FSD is better than a person, so I reduced that to say "as good".
The "problems with Tesla's data" as pointed out in the ars technica article doesn't make a comparison to an average driver, or give details on the number of miles driven per accident, etc etc
So, how are we to know where a regular human driver falls on the spectrum.
What data has Tesla provided about FSD? The only data I'm aware of is about Autopilot, and that data doesn't look very good when correctly interpreted.
It's not like they're deathtraps (it seems like mechanically/physically they are safe cars), and Tesla has repeatedly shown that they can disable and remove vehicle behaviour without a car going into a dealer/mechanic. So if it _is_ considered too unsafe a recall isn't needed.
In this case, a recall is very much needed, and would be issued OTA to all the Teslas. It would 100% be a recall, and none of the cars would go to a dealership.
I guess the Chevrolet Corvair is not so distant in human memory that the guy who made news dunking on it can still make a headline.
Haven’t heard of Nader in a long, long time time though. Maybe they’re dragging him out to split the vote again? Gotta start early for the presidential elections these days.
The car is continuously reminding you to keep your hands on the wheel with a prompt. You cannot ignore it without deliberately fooling the mechanism somehow. Ultimately it is your responsibility to hit the brakes and steer or stop the car. FSD is not a license to sit back and read a book. Without more information I find it hard to blame anyone but the driver. Are there cases of the car having a mind of its own and overriding the driver's command?
There was maybe some roadkill on the side of the road? Nothing to actually dodge - definitely a false positive.
My dad (a first-time Tesla driver) promptly turned off the assistance and we all had a better time.
Given then that the F stands for "full", this seems like a simple case of fraud to me.
Additionally, any automatic emergency braking functionality that a car has needs to be enabled by default, not something the driver must manually activate along with other features that are less broadly usable.