There is no version of FSD that doesn't require an attentive driver. You keep circling back to defining the safety of FSD itself. It is safe enough to release because it requires an attentive driver. If is obviously not safe enough to release in a form that absolves the driver of responsibility yet.
I’m not saying it doesn’t require an attentive driver. I am saying that it’s not always going to have an attentive driver. And even when it does, it will still make mistakes that even the attentive driver can’t get out of. I am saying that this is necessary to make the improvements that are necessary not only to make it better/safer, but also to make it so that it in the future FSD doesn’t require an attentive driver or any driver at all…You know, like actual FSD!
I'm not fanatical about FSD, I just believe in the data that shows FSD plus a human driver is already safer than a human driver alone.
I'm not even remotely fanatical about FSD, I rarely use it. I understand exactly what you're saying, but I'll still argue that FSD with an inattentive driver is still safer than that inattentive driver tooling down the road with their face in their phone. Similarly, a drunk passing out with FSD engaged is safer than a drunk passing out without it. Actively defeating FSDs attention checks is another issue entirely however.
I respect your opinion that it‘s already safer, but the data that’s been shown to us is partial at best. It‘s necessary to look much more granularly that just saying FSD + Human has X number of accidents vs Human Alone has Y and X < Y therefore FSD + Human = safer. That would be true if FSD was ubiquitous and driven over similar a number of miles. Instead there are many orders of magnitudes of difference and the amount of variability in the Human Alone population of drivers dwarfs the FSD + Human population. If we aren’t able to account for that then the data cannot be trusted to be an accurate representation of reality.
Here is an article talking about it. Their conclusion is that people switching from Gas cars to Tesla make 14.5% more claims in the first year. There isnt any mention of FSD or auto pilot. Hertz also listed increased crashes as a reason for dropping Tesla's from their fleet. I think its is more about the learning curve of instant torque and regen breaking. After reading into it, I do not think this has any impact on this FSD discussion.
Yep, that just illustrates how bad of drivers many humans are, they can't even switch to a different car without crashing more. The same principle applied, just to a lesser degree, when switching from one ICE car to another ICE car.
Operating a moving vehicle is complicated, even for FSD. You’ll notice recently all the hype about Ukraine getting F-16s. A week later 1 crashed during a mission. Are Ukrainians bad pilots? No. But learning a new platform takes time whether that’s an F-16, Model 3, or even whatever the next Dodge truck will be.
There's enough Teslas with enough miles to say FSD + Attentive driver is much safer and crashes less than other vehicles. There's over a billion miles and millions of Teslas.
I’m afraid that’s not how data science works. Tesla’s do not have the volumes or diversity of use cases needed to say that Yet. Still a way to go.
