> “The Full Self-Driving improvement will come as a quantum leap, because it’s a fundamental architectural rewrite, not an incremental tweak. I drive the bleeding edge alpha build in my car personally. Almost at zero interventions between home & work. Limited public release in 6 to 10 weeks.”
The degree to which he is willing to let the public test this software is disturbing. Sure, on the highway it may be safer than the average human. But going from any sort of alpha to a public release in 6-10 weeks on something this dangerous is nothing short of insane.
Musk makes me nervous that he and his team are going to push too far and too fast with putting self-driving in the hands of real users and the ensuing disaster/public outcry is going to be cataclysmic to the self-driving field. The whole "automatically have your car come to where you are in the parking lot" was a great example of an obvious disaster in the making that they made public anyway. I hope I'm wrong.
The worst is all the other people on the roads and sidewalks that might pay the price for rich people gadgetry and Silicon Valley move fast and kill people.
~33k people a year (About 90 per day) die in car accidents in the US because of humans. It’s us who have been given the long rope
There have been four deaths while Autopilot was engaged [1], almost all because the drivers weren’t paying attention. Autopilot has been used for 3 billion miles, 1 billion with Navigate on Autopilot (automatic passing and interchange routing), and 200,000 automated lane changes [2]. Is this not impressive?
Yes, self driving is hard. Yes, there will be fatalities due to edge cases. It is unreasonable to confine the tech to a test track until it’s perfect, considering the human incurred death rate as it stands. We should be racing to take the person out of the driver’s seat.
I tend to agree with everything you've said... however...
The cold-hard-reality is that if a self-driving car of any make were to careen into a school-bus full of children and burst into flames all automakers would very likely face severe restrictions going forward.
You know, for a long, long time Apple users claimed that Macs were unhackable. That was part of the hype and appeal for a long time. Problem was, Macs didn't reach critical mass in market penetration during those "unhackable years". Once there were enough in the general public to make it worth hacking, people start noticing/realizing that Macs weren't special, just irrelevant for a long time.
Same goes for self-driving cars. Critical mass has not been reached, not by a long shot. Most miles driven by autopilot cars have been supervised by either trained or very attentive drivers or even in very controlled environments. In other words, not how autopilot will be used in the real world what so ever. Because that's not the point. Either the car drives by itself or the person is driving. Not half here, half there. This is where the fear and hesitation exists. Will autopilot be inevitable? Probably. But Silicon Valley's cult like attitude to "break things fast" is not the attitude I want behind the software that is going to drive me to work or make me headless like the dude that crashed into a semitruck in Florida. At 33, I'm perfectly good at driving and will place my own life in my hands if "break things fast" is going to be their little bonfire chant. Someone else's idea of "the greater good" doesn't always live up to expectations.
> Yes, there will be fatalities due to edge cases.
If the public thinks this is how self-driving car work, I think self-driving will have a very hard time becoming popular and avoiding extreme regulation.
Who should be at fault in accidents? Tesla? The owner of the car? And if your death rate is correct and they will scale you would still have around 30 deaths every day. Automated driving is not the end all. I don't think we should put it into wide usage hastely for highway driving it may work, but it gets a lot worse as soon as you are off the high way I would believe.
The stats you give for human-driven deaths are for all conditions and all roads. Autopilot does not work in all conditions and all roads, so it’s not really a fair comparison.
Further to that, human deaths will be skewed towards drunk drivers and other “bad” drivers. Autopilot deaths will be “random”, so public perception will be worse even if the figures were exactly the same.
The other problem is complacency - the Apple software engineer had 2 seconds from his Tesla leaving the road to it hitting the gore point - and this was a bug he was aware of. If the car looks like it’s 100% infallible then peoples attention will naturally waver. How long does it take to react to a situation? Feel the weight of the steering again? Calculate the appropriate correction?
I think the system that replaces human drivers has to be 100 times better in terms of fatalities in order to be accepted by society.
Although I am concerned about safety and deeply skeptical of the Tesla self-driving claims, if the stats GP presented are to be believed at face value then a lot of your points are irrelevant. While society may not accept self driving until it is 100x safer it would be to their own deadly detriment... human beings are extremely bad at rationally managing risk.
Plus, it seems like the fewer interventions are needed without being zero, the more dangerous the system might be. I can see being lulled into a false sense of security if you have to do nothing most of the time - it's the times you need to do something (especially quickly) that you won't be paying attention and you end up under the semi or in the barrier.
Exciting, but literally "bleeding edge" stuff here.
This is why I think Waymo got it right that they were never going to market with level 3 self driving. You can't reasonably expect people to pay attention all the time when it works on its own 99.7% of the time.
I find it interesting that people equate/equated Elon Musk with Tony Stark. Years in the future, I think there’s a good chance people will look back and instead recognize him as an incredibly talented salesperson.
Years ago when he acquired a cult of personality and could do no wrong, I knew it was going to swing the other way.
I never subscribed to either view. He's a brilliant engineer and a talented businessman with some apparently significant personality flaws or emotional regulation issues. In other words he's human.
The problem with alot of people isn't the emotional regulation; it's the way he runs his businesses. He embraces a fake-it-till-you make it mindset to a huge degree. He's done so many things that are either fraud, near-fraud, or completely dishonest that it makes it impossible to support someone like that in his position.
It isn't very interesting. Elon is an attention seeking narcissist. He looks for it. When he runs around Twitter calling people paedophiles he's going to have opinions formed about him. What other outcome would you expect?
Elon's latest pronouncement is that Americans are complacent and entitled, especially Californians and New Yorkers. It's not exactly a roadmap to having people think highly of him.
As an engineer, I see him as an incredibly talented engineer that is also quite good at sales and marketing, but mostly because he talks about stuff like an engineer.
Would most companies even hire him for a salesperson role? Probably not if they've watched his presentations.
Do you really think Musk had anything to do with engineering design of those rockets? You really think he contributed to the rocket stabilization system or something?
I do actually. The level at which he can describe each low-level design decision shows he has a deep understanding of rocketry, chemistry, and material science and acts as more of a chief engineer rather than CEO.
Musk says he's doing engineering 80% of the time. If you watch him discuss the products he knows what he's talking about, he's intimately involved - he's not just the standard talking head CEO role.
Arguably he's already comparable, no? I mean he hasn't made a film yet, he also didn't inherit and spend his family's fortune - although yes, he had access to enough resources to build on his own efforts, successes.
i dont know the legal barriers or if some one tried it before, but self driving cars should train their model in Indian traffic with a human supervision , the traffic is very unpredictable and mostly experienced humans who have driven there long has nailed how to drive there, it will give some interesting data points
What is the legality of using an alpha quality build of autonomous driving software on a public road? The other drivers on the road obviously haven’t given their consent.
If I could prove I was on the road when he was and he passed me while his alpha software was running could I get a lawyer and sue?
51 comments
[ 3.5 ms ] story [ 88.2 ms ] threadThe degree to which he is willing to let the public test this software is disturbing. Sure, on the highway it may be safer than the average human. But going from any sort of alpha to a public release in 6-10 weeks on something this dangerous is nothing short of insane.
Still, we should expect autonomous vehicles to cope with that, which is certainly not the case yet.
Tesla has been given an incredibly long rope by regulators and it'll just take a really terrible accident or two to bring it all to a screeching halt.
It's hard to imagine Elon not realizing this, but the incessant drive to the endgame may cloud his judgement.
There have been four deaths while Autopilot was engaged [1], almost all because the drivers weren’t paying attention. Autopilot has been used for 3 billion miles, 1 billion with Navigate on Autopilot (automatic passing and interchange routing), and 200,000 automated lane changes [2]. Is this not impressive?
Yes, self driving is hard. Yes, there will be fatalities due to edge cases. It is unreasonable to confine the tech to a test track until it’s perfect, considering the human incurred death rate as it stands. We should be racing to take the person out of the driver’s seat.
[1] https://en.wikipedia.org/wiki/List_of_self-driving_car_fatal...
[2] https://electrek.co/2020/04/22/tesla-autopilot-data-3-billio...
The cold-hard-reality is that if a self-driving car of any make were to careen into a school-bus full of children and burst into flames all automakers would very likely face severe restrictions going forward.
Same goes for self-driving cars. Critical mass has not been reached, not by a long shot. Most miles driven by autopilot cars have been supervised by either trained or very attentive drivers or even in very controlled environments. In other words, not how autopilot will be used in the real world what so ever. Because that's not the point. Either the car drives by itself or the person is driving. Not half here, half there. This is where the fear and hesitation exists. Will autopilot be inevitable? Probably. But Silicon Valley's cult like attitude to "break things fast" is not the attitude I want behind the software that is going to drive me to work or make me headless like the dude that crashed into a semitruck in Florida. At 33, I'm perfectly good at driving and will place my own life in my hands if "break things fast" is going to be their little bonfire chant. Someone else's idea of "the greater good" doesn't always live up to expectations.
If the public thinks this is how self-driving car work, I think self-driving will have a very hard time becoming popular and avoiding extreme regulation.
Further to that, human deaths will be skewed towards drunk drivers and other “bad” drivers. Autopilot deaths will be “random”, so public perception will be worse even if the figures were exactly the same.
The other problem is complacency - the Apple software engineer had 2 seconds from his Tesla leaving the road to it hitting the gore point - and this was a bug he was aware of. If the car looks like it’s 100% infallible then peoples attention will naturally waver. How long does it take to react to a situation? Feel the weight of the steering again? Calculate the appropriate correction?
I think the system that replaces human drivers has to be 100 times better in terms of fatalities in order to be accepted by society.
"Almost" doing a lot of work here.
Exciting, but literally "bleeding edge" stuff here.
It's what leads to cult of personality or deep hatred.
I never subscribed to either view. He's a brilliant engineer and a talented businessman with some apparently significant personality flaws or emotional regulation issues. In other words he's human.
Elon's latest pronouncement is that Americans are complacent and entitled, especially Californians and New Yorkers. It's not exactly a roadmap to having people think highly of him.
Would most companies even hire him for a salesperson role? Probably not if they've watched his presentations.
If I could prove I was on the road when he was and he passed me while his alpha software was running could I get a lawyer and sue?