Tesla said it didn't have key data in a fatal crash, then a hacker found it

The marketing doesn't even matter. It either needs to be full self driving, or nothing at all. The "semi self-driving but you're still responsible when shit hits the fan" just doesn't work.

Humans are simply incapable of paying attention to a task for long periods if it doesn't involve some kind of interactive feedback. You can't ask someone to watch paint dry while simultaneously expect them to have < 0.5sec reaction time to a sudden impulse three hours into the drying process.

I have a SAE level 2 car. Those features DO help!

Same. Not having to worry about keeping the car between the lines allows me to keep my focus on the other cars around me more. Offloading the cognitive load of fine tuning allows more dedication to the bigger picture.

Framing is crucial. Example, why was the Autonomous Emergency Braking configured to brake violently to a full stop? Lets consider two scenarios, in both cases we're not paying enough attention to the outside world and are about to strike a child on a bicycle but the AEB policy varies.

1. AEB brakes violently to a full stop. We experience shock and dismay. What happened? Oh, a kid on a bike I didn't see. I nearly fucked up bad, good job AEB

2. AEB smoothly slows the vehicle to prevent striking the bicycle, we gradually become aware of the bike and believe we had always known it was there and our decision eliminated risk, why even bother with stupid computer systems?

Humans are really bad at accepting that they fucked up, if you give them an opportunity to re-frame their experience as "I'm great, nothing could have gone wrong" that's what they prefer, so, to deliver the effective safety improvements you need to be firm about what happened and why it worked out OK.

This makes no sense to me. Driving involves all senses, not just vision - if you're not feeling what the car is doing because you're not engaged with the steering wheel what good is it to see what's around you? I also don't understand how one has trouble staying between the lines with minimal cognitive input after more than a few months of driving.

Oh! And also, moving within the lane is sometimes important for getting a better look at what's up ahead or behind you or expressing car "body language" that allows others to know you're probably going to change lanes soon.

I don't have personal experience but friends with personal experience have sort of shifted my thinking on the topic. They'll note they do need to stay engaged but that it is genuinely useful on long drives in particular. The control handover is definitely an issue but so is manual driving in general. Their consensus is that the current state of the art is by no means perfect but it is improved and it's not like there aren't problems with existing manual driving even with some assistive systems.

I used to think this, but then I got a Model 3. I believe that FSD is presently better than most humans driving today even when they are theoretically “fully engaged in manual driving”.

FSD doesn’t lull humans into a false sense of security, humans do. FSD doesn’t let you use your phone while it’s on. This alone is an upgrade over most human beings, who think occasional quick phone usage while driving is fine (at least for themselves).

I believe that if you replaced all human drivers in the US with FSD as it exists today, fatalities would go down immediately.

Humans are not a gold standard, and the current median human driver is easy to outperform on safety.

My car requires hands on the wheel to continue to operate. So I do feel it moving.

> I also don't understand how one has trouble staying between the lines with minimal cognitive input after more than a few months of driving.

Once you have something assist you with that, you'll notice how much "effort" you are actually putting towards it.

I drive a VW with lane-keep assist and adaptive cruise control and automatic emergency braking. It won't change lanes for me, but aside from the requirements that I have my hands on the wheel, could otherwise drive itself on the highway.

I commute mainly on the highway about 45-1hr each way every day and it makes a big difference for driver fatigue. I was honestly a bit surprised. Even though, I'm steering, it requires less effort. I don't have my foot on the gas and I'm not having to adjust my speed constantly.

Critically, though, I do have to pay attention to my surroundings. It's not taking so much out of my driving that I can't stay engaged to what's happening around me.

So this is a really good example of small sample size intuition being a big challenge. Fatalities happen on the order of billion miles driven - obviously people don’t come to that. Take a few thousand miles of positive experience sets a statistical floor on accident rates, but that is orders of magnitude away from how safe (or unsafe, depending on how you look at it) human drivers are on average. FSD and other, less capable L2 systems are amazing at paying attention in situations where humans fail, but also tend to have major limitations in places humans will largely do great most of the time. Your experience, as positive as it has been, doesn’t support the assertion that fatalities would decrease.

I see median human drivers all of the time, and I see median FSD all of the time. I don’t need to drive a billion miles to have a valid opinion that one is better than the other.

I’m sure things are very different out around the edges, as you note, but the majority of the time humans in cars kill people it isn’t because they were in an edge case - quite the opposite. They were just driving home from the bar like they do every night.