Who the hell thinks beta software is appropriate for real-world applications in something as dangerous as vehicle control at highway speeds?
I’ve come to believe that all Teslas should be recalled until they get their act together. They’re getting people hurt and killed by field testing their experiments on roadways that we paid for.
I agree. However, I also acknowledge that with the US’s legal fiction that “Corporations are people,” it’s unlikely that any CEO will ever be held personally responsible for anything except failing to make enough profit for greedy moneygrubbers.
That’s not true. Corporations can’t commit crimes because they are just legal entities. People commit crimes.
Your main issue is that many things that you think are crimes (dumping waste, not paying employees, stiffing suppliers, accidents) are not crimes. They are civil or regulatory issues. If you care, you should pick one and lobby your state to make it a crime.
[M]any things that you think are crimes (dumping waste, not paying employees, stiffing suppliers, accidents) are not crimes.
Again, the US federal and state governments would like a word. There’s not a locale in the country that considers theft and intentional illegal pollution not to be criminal acts with fines and, for natural person, imprisonment as punishments. However, you are correct that accidents are generally not considered crimes, although penalties may still apply if they were the result of carelessness or neglect.
Well, it’s a beta for “full self driving” and it’s really quite easy to argue that it even is capable of doing that. It’s just not really close to the required certainty.
It propably acts fine in 99%+ of decisions. It’s just such a bitch to get that last percent filled in as much as required
I’ll clarify, I don’t mind it being called FSD beta because that is the end goal, but Elon has tweeted about FSD being “released” and very public accidents have occurred shortly afterwards. He should be adding disclaimers all over, but instead he makes it sound like it’s totally safe.
While I agree, let’s not pretend like this is limited just to Tesla. My feed lately has had numerous stories of crazy FSD taxis as well.
I also have to say that one of my concerns with FSD is the deterioration in people’s driving skills and their awareness of their car’s abilities (especially as those change over time). Leaving aside all the wisecracks about people’s normal abilities or not paying attention anyway, let’s take a snowstorm. FSD can’t drive in it, so you’re left with regular human drivers going manual in their cars. But they haven’t actually driven themselves in a while, so they’ve forgotten some of the lessons they learned like how to apply the brakes differently in ice and snow, they don’t know where the corners of their car are, they’re driving entirely too fast and - because their FSD car was compensating for mechanical issues - they’re not aware that their tires are near-bald and the brakes are iffy.
Thing is, I know this is something that’s going to happen. I just don’t know how we can mitigate the risks.
IMHO, Waymo a Cruise AVs are different animals. They have LiDAR. Musk is still hell bent on developing a camera-only system, which is inferior. But it’s cheaper and less bulky, so Musk is all about it.
It seems that both Waymo and Cruise are more likely to already surpass average human driving safety, than not.
I’m really curious on how the next FSD version (which apparently completely relies on neural nets for driving) play out.
Not that I think it’ll be particularily good, just particularily interesting.
Who the hell thinks beta software is appropriate for real-world applications in something as dangerous as vehicle control at highway speeds?
I honestly think it’s a mixture of public perception and liability. The company can try to spin negative insurances as “well, we said it’s still in beta, there are bugs we’re still ironing out”. And legally I think the stance is “we said it’s a beta version, if you used it in a dangerous situation that’s on you”.
I know it doesn’t exactly work that way, but I genuinely think they’re positioning that way at that if (read: when) a legal case pops up they can use the “beta” moniker as part of a defence.
Yeah my dude, but that’s not the part I’m disagreeing with. You made one false claim and one true one, and I’m pointing out that the false one is false.
Who the hell thinks beta software is appropriate for real-world applications in something as dangerous as vehicle control at highway speeds?
I’ve come to believe that all Teslas should be recalled until they get their act together. They’re getting people hurt and killed by field testing their experiments on roadways that we paid for.
More importantly, hold musk responsible for the mayhem. They call it “full self driving” when it has not qualified to be called that.
I agree. However, I also acknowledge that with the US’s legal fiction that “Corporations are people,” it’s unlikely that any CEO will ever be held personally responsible for anything except failing to make enough profit for greedy moneygrubbers.
Corporate death penalty. Revoke their corporate charter. No more company.
While I personally agree, under US shareholder-primacy laws, this would likely be impossible in the current era.
That’s not true. Corporations can’t commit crimes because they are just legal entities. People commit crimes.
Your main issue is that many things that you think are crimes (dumping waste, not paying employees, stiffing suppliers, accidents) are not crimes. They are civil or regulatory issues. If you care, you should pick one and lobby your state to make it a crime.
The US Supreme Court begs to differ:
The list goes on.
Again, the US federal and state governments would like a word. There’s not a locale in the country that considers theft and intentional illegal pollution not to be criminal acts with fines and, for natural person, imprisonment as punishments. However, you are correct that accidents are generally not considered crimes, although penalties may still apply if they were the result of carelessness or neglect.
Yeah that’s straight fraud. FTC should penalize them for every fsd car sold. And make them liable for each crash when it was enabled.
Well, it’s a beta for “full self driving” and it’s really quite easy to argue that it even is capable of doing that. It’s just not really close to the required certainty. It propably acts fine in 99%+ of decisions. It’s just such a bitch to get that last percent filled in as much as required
I’ll clarify, I don’t mind it being called FSD beta because that is the end goal, but Elon has tweeted about FSD being “released” and very public accidents have occurred shortly afterwards. He should be adding disclaimers all over, but instead he makes it sound like it’s totally safe.
While I agree, let’s not pretend like this is limited just to Tesla. My feed lately has had numerous stories of crazy FSD taxis as well.
I also have to say that one of my concerns with FSD is the deterioration in people’s driving skills and their awareness of their car’s abilities (especially as those change over time). Leaving aside all the wisecracks about people’s normal abilities or not paying attention anyway, let’s take a snowstorm. FSD can’t drive in it, so you’re left with regular human drivers going manual in their cars. But they haven’t actually driven themselves in a while, so they’ve forgotten some of the lessons they learned like how to apply the brakes differently in ice and snow, they don’t know where the corners of their car are, they’re driving entirely too fast and - because their FSD car was compensating for mechanical issues - they’re not aware that their tires are near-bald and the brakes are iffy.
Thing is, I know this is something that’s going to happen. I just don’t know how we can mitigate the risks.
IMHO, Waymo a Cruise AVs are different animals. They have LiDAR. Musk is still hell bent on developing a camera-only system, which is inferior. But it’s cheaper and less bulky, so Musk is all about it.
Oh, I completely agree on all points. None of them are ready for full autopilot.
https://arstechnica.com/cars/2023/09/are-self-driving-cars-already-safer-than-human-drivers/
It seems that both Waymo and Cruise are more likely to already surpass average human driving safety, than not.
I’m really curious on how the next FSD version (which apparently completely relies on neural nets for driving) play out.
Not that I think it’ll be particularily good, just particularily interesting.
I honestly think it’s a mixture of public perception and liability. The company can try to spin negative insurances as “well, we said it’s still in beta, there are bugs we’re still ironing out”. And legally I think the stance is “we said it’s a beta version, if you used it in a dangerous situation that’s on you”.
I know it doesn’t exactly work that way, but I genuinely think they’re positioning that way at that if (read: when) a legal case pops up they can use the “beta” moniker as part of a defence.
Agreed. I’m just not sure how regulators justify allowing software that claims to be beta to operate a vehicle autonomously.
Users do. See all the people running iOS betas on their daily drivers, then complaining about a forced reset or data loss.
A cell phone is not a motor vehicle.
deleted by creator
That’s not true. It’s been open to anyone who requests it for many months
deleted by creator
Inside EVs article:
Tweet from Elon dated November 24, 2022:
deleted by creator
Yeah my dude, but that’s not the part I’m disagreeing with. You made one false claim and one true one, and I’m pointing out that the false one is false.
deleted by creator