"This is not good," Musk said on an earnings call this week
Meaning that people are intentionally abusing the system. Not that the system is inherently unsafe or that we are marketing it wrongly. The headline is 'People doing ‘crazy things’ with Tesla’s autopilot are spoiling it for everybody'. The people who are eventually ruining it are/is you! A few idiots killing themselves will cause other idiots like yourself to panic and pull a useful technology that saves lives.
That's not some clueless idiot; that's a paid test driver who was trained to do his job. Human nature.
A person who took a risk, like millions of people every day. What could you possibly think this proves other than that cars should be banned outright until there are no more traffic fatalities? Please tell me you have never changed a radio station, reached in the backseat for a water bottle or some other action that momentarily took your full attention off the road. Then I can completely disregard everything you say, as opposed to thinking you are just a well meaning alarmist with no practical math skills.
I'd pull the feature until it can drive itself.
So you would pull a feature that saves lives, that causes fewer crashes, less property damage because...? Demonstrate that the feature will cause more deaths by being enabled than if it were disabled. Then you would have a serious point. Until then it is just fear mongering, luddite bullshit.
Reassuring point #1: even after learning the road it will (almost) never do something spectacularly stupid if conditions are now different.
Reassuring point #2: autonomous cars can quickly communicate to others changes in the environment (for now this is very slow, but it will get steadily faster)
Reassuring point #3: when there is a critical mass of autonomous cars such changes will either a) be sent out as alerts by those construction workers, or b) marked in a new clear way that is easily identifiable to autonomous cars (could be as simple as a single sign "ALERT!" that autonomous cars know to throw out their optimized routes and be extra cautious)
Reassuring point #4: humans are spectacularly bad at this and even the occasional computer failure will be much better than the status quo. A human who has driven down the same road every day for the last year is often going to not notice a new stop sign. Further, every other human will make that same mistake as their communication system is far slower.
Okay, I read another of the articles that you think proves your point. You quote the summary:
The feature itself has gained a lot of fame in the recent months thanks to its obvious novelty value and the fact that it is the first hands-off, self-driving technology on the market today.
Except the actual article goes into much more detail to explain what that means:
The second milestone is Semi-Autonomous driving, something only Tesla can claim at the moment, and consists of the car driving itself (hands-off the steering wheel) with the driver being a necessary requirement for regular monitoring. In this case, the car will handover control to the human in various scenarios. The element of the human driver is assumed to be an active participant in the process – albeit one which doesn’t interfere for some (if not most) of the time. Basically, if you crash the car while on Autopilot – you are responsible.
It is absolutely clear in that article that the autopilot works as Tesla officially claims and that any reasonable person would expect. Any person who bought a Tesla thinking it would let them take a nap while driving from SF to LA is either insane or a pathological liar. There are certainly thousands of people who have only vague ideas and think Teslas are self-driving, these people don't own them and are not serious customers. Most people are superficial idiots, what are you going to do?
You are holding Tesla responsible for every tweet that could possibly be misinterpreted. It's a f*king tweet! It has to be short and almost always is a bastardized executive summary. That is like holding VW responsible for tweets that say "the new Porsche has 500 horsepower and can go 195". Good lord man, why are they telling people to go 195 mph on a public highway!
Have you test-driven a Tesla? Do it, test the auto pilot and ask your "advisor" about it. I guarantee he will not mislead you into thinking you are getting a self-driving car. At best you can manipulate him into saying "I guess you could play Jenga while going down the highway, but we would not recommend that and it would be a seriously dumb thing to do."
If you read "between" the lines and take statements out of context you can make their press conform to any preconceived notion that you want to. Which is exactly what you did. The first example I looked at, the gizmodo reference shows to what great lengths you'll go to to exaggerate your point. The full quote from Musk is:
The probability of having an accident is 50% lower if you have Autopilot on. Even with our first version. So we can see basically what’s the average number of kilometers to an accident – accident defined by airbag deployment. Even with this early version, it’s almost twice as good as a person.
It is absolutely clear that he is talking about autopilot on with a reasonable driver behind the wheel. He is definitely not talking about how well the system works if the driver is asleep or non-existant.
The author of this article you seem to be so inflamed about goes on to mention all the reasonable caveats that Tesla (or I) would make. No reasonable person would read that article and assume that Musk or the author thought you could drive completely hands off. Did you even read the whole article or did you just google for random one sentence quotes that could be twisted to support your argument.
Every Tesla employee and on every test drive they have stressed that it is NOT hands free, set it and forget it, driving. I don't know what your grievance is against Tesla, but you are completely wrong.
Do you have an example of such marketing? I've never seen it.
There is no data to suggest that Tesla is putting the general public at increased risk. It may shift the risk to idiots who abuse the system, but the average sensible driver might very well be better off. Adding too much of a deterrent might actually cost lives because it will limit how often sensible people engage the system.
I wish there were a legal way for me to make bets against overconfident drivers such as this. If there were, these perfect drivers would shut up pretty quickly (or go bankrupt pretty quickly).
Every Tesla employee I've talked to and on every test drive they've made it very clear that the technology is limited to certain situations and the driver has to pay attention and maintain control. They still gush over how advanced (and cool) autopilot is, but they don't make absurd claims, IME.
Self-driving cars need a couple of things before they become safe
Who says they're not safe now? You're going to have to define "safe" a lot more explicitly if you're going to make demands on what technology needs to be in place before we deploy it. If you mean perfectly safe, then we will never get there. So why isn't the bar just safer than current human drivers under normal conditions?
1. An effective deadman switch that will assure that the car will come to a safe stop if the "driver" is not paying attention
If the goal is just to be safer then why wouldn't you make this a requirement for all cars? You don't need special rules for self-driving or autopilot cars.
2. Software that can handle the ethical questions that arise constantly when driving.
These are not ethical questions, they are purely technological and statistical problems. You have a hard time wrapping your head around how different such technology is compared to humans. We don't program the software to understand Bavarian festival rituals, we program it to detect that there are a lot more obstacles on this road today, some of which are moving erratically. The speed limit is irrelevant to a self-driving car, the only requirement is that it goes under it (for now). If a self-driving car can navigate such a busy street at 25 with the same safety as you when you have slowed down to 15, then why force it to go 15 as well?
If you are only going 3 blocks it might be more efficient to walk than to take a train.
If you are going from New York to L.A. it might be faster to take a plane than a bus or a train.
The point is that you made an unqualified statement that public transportation is more efficient than a single person in a car. That may be generally true for the city where you live, but it is not a universal truth everywhere. I would bet that it's not even always true where you live; I've been in countless cities throughout the world and there are always times and routes where it is more efficient to drive.
So please, don't disparage the usefulness of self-driving cars or driving in general. Let people pay the fair cost for their travel and let them choose whether they would rather be driven door to door or deal with public transportation or even some other option. Not everyone is exactly like you.
Look, you can make absurdly strict definitions to suit your predisposed obnoxious opinions, but when the conclusions that result are so "obviously" absurd you demonstrate that you have no idea what you're talking about and are just being an ass.
My car has a number of safety/convenience features such as adaptive cruise control, lane change warning indicator, parking assist and proximity indicators, emergency crash avoidance,.... Every single one of those has edge cases and limitations that I am not fully aware of. Even if they don't know exactly which cars are being detected, it is "obvious" to most people how they are used and generally what they are doing. I cannot just change lanes whenever my warning indicator is not lit.
By your ridiculous standard every one of those features should be yanked from the vehicle until they are complete. Your absurd semantic torture leads to ridiculous conclusions, which means you're just being a dick. Bye.
No, the "given that" refers to "given that we will never have complete information." There are practical limitations to what the technology can do and practical assumptions that people can make.
The only way to leave out unfinished work from cars is to never add such features to cars. They will never, ever be complete, they will always be making incremental improvements. Just about every new luxury car has some such features that do not live up to the standard of perfection that you demand because of your obstinance. Obstinance that would cause lives to be lost if you were in charge.
As I just said, it is necessarily incomplete. You are demanding the impossible and throwing out what might be a life saving technology because it doesn't meet your level of perfection.
Why does it matter if some abuse the autopilot system if it safer overall? If there were evidence that autopilot caused more damage than it prevented it would be quite reasonable to delay the technology (by legal order if necessary), and I am reasonably sure that if Elon sincerely believed that it was killing people he would.
If Google acts as you claim then it is a financial decision and not an ethical decision. If you do not release a product that will save lives just because it isn't as safe as it possibly could be, then you might in fact be considered evil.
That is where the GP's leap of logic fails. People losing attention makes the system less safe than it would be if people could somehow be forced to focus. It does not necessarily make the system less safe than a car that has no autopilot feature at all.
If self driving cars can't save enough lives (and property) to offset the costs of the vehicle then they will have failed. There will still be a market for those who want to avoid driving, but that will be a much smaller market. Luddites, er I mean "driving enthusiasts" should still be able to drive themselves as long as they are willing to pay the cost of the risk they add.
Even a $50k upcharge for such technology would still be useful for taxis, delivery vehicles and others who rack up a lot of miles.
The other options are only more "efficient" if you place no value on your time or personal convenience. You might choose to spend an hour on public transportation instead of a 30 minute drive to save $10, but I would usually not. So just put a fair price on everything and let people choose what they value more, rather than dictating how efficient others should be.
Autopilot gives feedback on the objects it detects, but of course that information, and what it plans to do about them, is not complete. If a human couldn't understand everything it is doing all the time we wouldn't need the computer.
Given that, I think it is obvious what trivial parts it is handling. I wouldn't expect it to gracefully handle a vehicle crossing the road like this situation and not having to worry about the cars immediately around me makes it possible for me to keep an eye out for such situations. I'm not saying that every driver (or even a good one) WILL be looking a little deeper, but they don't have to for such technology to be safer than unassisted drivers.
Many collisions are avoided by a car taking evasive action to avoid hitting another car that makes a mistake, but a driver should never make an assumption on the magnitude that you are asking. It may be reasonable to switch lanes in front of another car that is going faster than you if you give them enough space to slow down; they don't have unlimited rights to the highway in front of them. This is very different situation.
Slashdot Mirror
"This is not good," Musk said on an earnings call this week
Meaning that people are intentionally abusing the system. Not that the system is inherently unsafe or that we are marketing it wrongly. The headline is 'People doing ‘crazy things’ with Tesla’s autopilot are spoiling it for everybody'. The people who are eventually ruining it are/is you! A few idiots killing themselves will cause other idiots like yourself to panic and pull a useful technology that saves lives.
That's not some clueless idiot; that's a paid test driver who was trained to do his job. Human nature.
A person who took a risk, like millions of people every day. What could you possibly think this proves other than that cars should be banned outright until there are no more traffic fatalities? Please tell me you have never changed a radio station, reached in the backseat for a water bottle or some other action that momentarily took your full attention off the road. Then I can completely disregard everything you say, as opposed to thinking you are just a well meaning alarmist with no practical math skills.
I'd pull the feature until it can drive itself.
So you would pull a feature that saves lives, that causes fewer crashes, less property damage because ...? Demonstrate that the feature will cause more deaths by being enabled than if it were disabled. Then you would have a serious point. Until then it is just fear mongering, luddite bullshit.
Reassuring point #1: even after learning the road it will (almost) never do something spectacularly stupid if conditions are now different.
Reassuring point #2: autonomous cars can quickly communicate to others changes in the environment (for now this is very slow, but it will get steadily faster)
Reassuring point #3: when there is a critical mass of autonomous cars such changes will either a) be sent out as alerts by those construction workers, or b) marked in a new clear way that is easily identifiable to autonomous cars (could be as simple as a single sign "ALERT!" that autonomous cars know to throw out their optimized routes and be extra cautious)
Reassuring point #4: humans are spectacularly bad at this and even the occasional computer failure will be much better than the status quo. A human who has driven down the same road every day for the last year is often going to not notice a new stop sign. Further, every other human will make that same mistake as their communication system is far slower.
The feature itself has gained a lot of fame in the recent months thanks to its obvious novelty value and the fact that it is the first hands-off, self-driving technology on the market today.
Except the actual article goes into much more detail to explain what that means:
The second milestone is Semi-Autonomous driving, something only Tesla can claim at the moment, and consists of the car driving itself (hands-off the steering wheel) with the driver being a necessary requirement for regular monitoring. In this case, the car will handover control to the human in various scenarios. The element of the human driver is assumed to be an active participant in the process – albeit one which doesn’t interfere for some (if not most) of the time. Basically, if you crash the car while on Autopilot – you are responsible.
It is absolutely clear in that article that the autopilot works as Tesla officially claims and that any reasonable person would expect. Any person who bought a Tesla thinking it would let them take a nap while driving from SF to LA is either insane or a pathological liar. There are certainly thousands of people who have only vague ideas and think Teslas are self-driving, these people don't own them and are not serious customers. Most people are superficial idiots, what are you going to do?
You are holding Tesla responsible for every tweet that could possibly be misinterpreted. It's a f*king tweet! It has to be short and almost always is a bastardized executive summary. That is like holding VW responsible for tweets that say "the new Porsche has 500 horsepower and can go 195". Good lord man, why are they telling people to go 195 mph on a public highway!
Have you test-driven a Tesla? Do it, test the auto pilot and ask your "advisor" about it. I guarantee he will not mislead you into thinking you are getting a self-driving car. At best you can manipulate him into saying "I guess you could play Jenga while going down the highway, but we would not recommend that and it would be a seriously dumb thing to do."
The probability of having an accident is 50% lower if you have Autopilot on. Even with our first version. So we can see basically what’s the average number of kilometers to an accident – accident defined by airbag deployment. Even with this early version, it’s almost twice as good as a person.
It is absolutely clear that he is talking about autopilot on with a reasonable driver behind the wheel. He is definitely not talking about how well the system works if the driver is asleep or non-existant.
The author of this article you seem to be so inflamed about goes on to mention all the reasonable caveats that Tesla (or I) would make. No reasonable person would read that article and assume that Musk or the author thought you could drive completely hands off. Did you even read the whole article or did you just google for random one sentence quotes that could be twisted to support your argument.
Every Tesla employee and on every test drive they have stressed that it is NOT hands free, set it and forget it, driving. I don't know what your grievance is against Tesla, but you are completely wrong.
Fortunately, reckless and selfish truck drivers (like fluffernutter's friends) are going to be the first to be replaced by self driving cars.
Do you have an example of such marketing? I've never seen it.
There is no data to suggest that Tesla is putting the general public at increased risk. It may shift the risk to idiots who abuse the system, but the average sensible driver might very well be better off. Adding too much of a deterrent might actually cost lives because it will limit how often sensible people engage the system.
Or you can take a reasonable middle ground like any sane person would do.
I wish there were a legal way for me to make bets against overconfident drivers such as this. If there were, these perfect drivers would shut up pretty quickly (or go bankrupt pretty quickly).
Every Tesla employee I've talked to and on every test drive they've made it very clear that the technology is limited to certain situations and the driver has to pay attention and maintain control. They still gush over how advanced (and cool) autopilot is, but they don't make absurd claims, IME.
Self-driving cars need a couple of things before they become safe
Who says they're not safe now? You're going to have to define "safe" a lot more explicitly if you're going to make demands on what technology needs to be in place before we deploy it. If you mean perfectly safe, then we will never get there. So why isn't the bar just safer than current human drivers under normal conditions?
1. An effective deadman switch that will assure that the car will come to a safe stop if the "driver" is not paying attention
If the goal is just to be safer then why wouldn't you make this a requirement for all cars? You don't need special rules for self-driving or autopilot cars.
2. Software that can handle the ethical questions that arise constantly when driving.
These are not ethical questions, they are purely technological and statistical problems. You have a hard time wrapping your head around how different such technology is compared to humans. We don't program the software to understand Bavarian festival rituals, we program it to detect that there are a lot more obstacles on this road today, some of which are moving erratically. The speed limit is irrelevant to a self-driving car, the only requirement is that it goes under it (for now). If a self-driving car can navigate such a busy street at 25 with the same safety as you when you have slowed down to 15, then why force it to go 15 as well?
If you are only going 3 blocks it might be more efficient to walk than to take a train.
If you are going from New York to L.A. it might be faster to take a plane than a bus or a train.
The point is that you made an unqualified statement that public transportation is more efficient than a single person in a car. That may be generally true for the city where you live, but it is not a universal truth everywhere. I would bet that it's not even always true where you live; I've been in countless cities throughout the world and there are always times and routes where it is more efficient to drive.
So please, don't disparage the usefulness of self-driving cars or driving in general. Let people pay the fair cost for their travel and let them choose whether they would rather be driven door to door or deal with public transportation or even some other option. Not everyone is exactly like you.
Look, you can make absurdly strict definitions to suit your predisposed obnoxious opinions, but when the conclusions that result are so "obviously" absurd you demonstrate that you have no idea what you're talking about and are just being an ass.
My car has a number of safety/convenience features such as adaptive cruise control, lane change warning indicator, parking assist and proximity indicators, emergency crash avoidance, .... Every single one of those has edge cases and limitations that I am not fully aware of. Even if they don't know exactly which cars are being detected, it is "obvious" to most people how they are used and generally what they are doing. I cannot just change lanes whenever my warning indicator is not lit.
By your ridiculous standard every one of those features should be yanked from the vehicle until they are complete. Your absurd semantic torture leads to ridiculous conclusions, which means you're just being a dick. Bye.
No, the "given that" refers to "given that we will never have complete information." There are practical limitations to what the technology can do and practical assumptions that people can make.
The only way to leave out unfinished work from cars is to never add such features to cars. They will never, ever be complete, they will always be making incremental improvements. Just about every new luxury car has some such features that do not live up to the standard of perfection that you demand because of your obstinance. Obstinance that would cause lives to be lost if you were in charge.
As I just said, it is necessarily incomplete. You are demanding the impossible and throwing out what might be a life saving technology because it doesn't meet your level of perfection.
Why does it matter if some abuse the autopilot system if it safer overall? If there were evidence that autopilot caused more damage than it prevented it would be quite reasonable to delay the technology (by legal order if necessary), and I am reasonably sure that if Elon sincerely believed that it was killing people he would.
If Google acts as you claim then it is a financial decision and not an ethical decision. If you do not release a product that will save lives just because it isn't as safe as it possibly could be, then you might in fact be considered evil.
That is where the GP's leap of logic fails. People losing attention makes the system less safe than it would be if people could somehow be forced to focus. It does not necessarily make the system less safe than a car that has no autopilot feature at all.
If you had a technology that saved lives, but wasn't perfect, and you did NOT release it then I would question your own ethics.
What can happen in 30 seconds that can't happen in 5 seconds?
If self driving cars can't save enough lives (and property) to offset the costs of the vehicle then they will have failed. There will still be a market for those who want to avoid driving, but that will be a much smaller market. Luddites, er I mean "driving enthusiasts" should still be able to drive themselves as long as they are willing to pay the cost of the risk they add.
Even a $50k upcharge for such technology would still be useful for taxis, delivery vehicles and others who rack up a lot of miles.
The other options are only more "efficient" if you place no value on your time or personal convenience. You might choose to spend an hour on public transportation instead of a 30 minute drive to save $10, but I would usually not. So just put a fair price on everything and let people choose what they value more, rather than dictating how efficient others should be.
Autopilot gives feedback on the objects it detects, but of course that information, and what it plans to do about them, is not complete. If a human couldn't understand everything it is doing all the time we wouldn't need the computer.
Given that, I think it is obvious what trivial parts it is handling. I wouldn't expect it to gracefully handle a vehicle crossing the road like this situation and not having to worry about the cars immediately around me makes it possible for me to keep an eye out for such situations. I'm not saying that every driver (or even a good one) WILL be looking a little deeper, but they don't have to for such technology to be safer than unassisted drivers.
Your logic is sound even if your math is lousy ;-)
Many collisions are avoided by a car taking evasive action to avoid hitting another car that makes a mistake, but a driver should never make an assumption on the magnitude that you are asking. It may be reasonable to switch lanes in front of another car that is going faster than you if you give them enough space to slow down; they don't have unlimited rights to the highway in front of them. This is very different situation.
For so many more reasons than just this, actually.