Tesla "Full Self Driving" not capable of autonomous driving

[Irace86.2.0]

Originally Posted by AnalogMan

How would people feel if someone sold oleander and named it "Miracle COVID Cure", or a bottle of vitamins called "Complete Cancer Cure"?

It sounds like your argument against the name Autopilot is that people are dumb, so if they call it ProPilot or Supercruze or anything else then people will not use their autonomous systems in careless ways. I fundamentally disagree, but even if that were the case, you are basing this on the idea that people have a preconceived definition of Autopilot that is also flawed:

https://en.wikipedia.org/wiki/Autopilot

An autopilot is a system used to control the trajectory of an aircraft, marine craft or spacecraft without requiring constant manual control by a human operator. Autopilots do not replace human operators. Instead, the autopilot assists the operator's control of the vehicle, allowing the operator to focus on broader aspects of operations (for example, monitoring the trajectory, weather and on-board systems).

Again, what would you call it if I could announce a destination, and a Tesla drives me to that location without any driver inputs? Would that not be Full Self Driving Mode? Would you rather they call it Practically-Almost-All-The-Time Full Self Driving Mode because I might need to intervene in an emergency or might need to make corrections?

Going back to your example above, this isn't false advertising because it does what it says it is doing, and it is the name of a mode or feature; it is capable of fully driving from A to B with no inputs. Exceptions don't negate the validity of something. Should a manufacture of a blood pressure medication be forbidden to claim or market their medication as being able to lower blood pressure if the medication doesn't work all the time or work on a small percentage of people? This is not how consumer protections work.

I watched the first video, and it is hard to know if a human moved the car or didn't move the car, or if the car went in the other lane, but then corrected itself or if a human corrected the car. It is also hard to tell if a human would have made a different move to avoid an object because I couldn't identify the object in some clips. I talked about the overturn truck example. Where was the driver's attention? Would the driver have been paying attention in any other vehicle, or would they have the car on cruise control and be distracted? I don't know. A failure in the system is also a failure of the driver, yet we are blaming the car? Statistically, if there are less deaths and less accidents using Autopilot than not using Autopilot then regardless of any outliers, we want people to use Autopilot. As the systems get better and better, people will push the car's abilities more and more, but that isn't on Tesla, especially in the Inside Edition video. We are moving to autonomous driving, and it was never going to just appear overnight in perfect form, so we should have expected to have idiots abusing the system, and we weren't disappointed. The next video shows many things, but you may have read the title and not watched the video because the semi crash wasn't using Autopilot because this guy makes that clear in his videos (I'm already subscribed), which he does make clear in one of the clips in the video where Autopilot veered to avoid getting side swiped when an Infiniti threaded between the Tesla and another car. I recall the case in the final video, and the wife said the husband on several occasions noted that the Autopilot failed at the same spot he died at, which tells me, he was either testing them with his life, or he was being complacent. Again, I fail to see how this is Tesla's fault.

[AnalogMan]

I'm all for developing and implementing self-driving technology. At some point, probably sooner than later, autonomous driving technology will, on the whole, be statistically safer than actual live people doing the driving, as long as it's used appropriately (though we shouldn't totally rely on technology to cure all of our human ills; history has shown that it usually doesn't work out well).

Someday, hopefully long after I'm gone, the world will be filled with completely autonomous, all-electric homogenous transportation pods (hopefully powered by a grid based on clean, renewable energy sources instead of burning stuff), efficiently moving shiny happy people to their destinations. It will be a cleaner and safer (though in my curmudgeon's eyes, more boring) world (I'm a socialist of the old school, so it's not my idea of utopia, but it's the 'future' and I probably won't be here to see it).

Put this technology in all Teslas. License it to Ford and GM so they can put it in their cars. Develop a retrofit of it for existing cars, if that's possible.

Just for the love of all the amorphous celestial deities of your choice call it something else. Name it Super-Duper Cruise. Name it Elon's Genius. Name it Musk's Brilliance. Name it Tesla's Terrific. Just don't call it 'Full Self Driving'. It's not just a 'name'. It creates expectations, and those become perceived as reality in people's minds. Too many people are too myopic to see it realistically for what it is, and what it isn't, and as a result people will needlessly die in the process of implementing an otherwise beneficial technology.

[Irace86.2.0]

Originally Posted by AnalogMan

I'm all for developing and implementing self-driving technology. At some point, probably sooner than later, autonomous driving technology will, on the whole, be statistically safer than actual live people doing the driving, as long as it's used appropriately (though we shouldn't totally rely on technology to cure all of our human ills; history has shown that it usually doesn't work out well).

Someday, hopefully long after I'm gone, the world will be filled with completely autonomous, all-electric homogenous transportation pods (hopefully powered by a grid based on clean, renewable energy sources instead of burning stuff), efficiently moving shiny happy people to their destinations. It will be a cleaner and safer (though in my curmudgeon's eyes, more boring) world (I'm a socialist of the old school, so it's not my idea of utopia, but it's the 'future' and I probably won't be here to see it).

Put this technology in all Teslas. License it to Ford and GM so they can put it in their cars. Develop a retrofit of it for existing cars, if that's possible.

Just for the love of all the amorphous celestial deities of your choice call it something else. Name it Super-Duper Cruise. Name it Elon's Genius. Name it Musk's Brilliance. Name it Tesla's Terrific. Just don't call it 'Full Self Driving'. It's not just a 'name'. It creates expectations, and those become perceived as reality in people's minds. Too many people are too myopic to see it realistically for what it is, and what it isn't, and as a result people will needlessly die in the process of implementing an otherwise beneficial technology.

I said before that I disagree with your fundamental premise that the name causes people to unknowingly lower their guard while driving. I think the name has nothing to do with it. I think the owners are very aware of what the system can and can't do, and I think people become complacent because they can, because the system allows them to be complacent. It doesn't matter that the screen may flash a warning or the vehicle chimes a warning if someone doesn't touch the steering wheel during Autopilot. People will choose to ignore those things because they can until they can't. Changing the name won't alter this behavior.

Tesla could alter this behavior by pulling the car over when the driver isn't looking forward with open eyes, or it could do that if someone doesn't touch the steering wheel once a minute or something pestering, but obviously, they don't want to force someone. Maybe some states will pass laws that require such things like how some states have laws against driving while talking/texting.

As far as safety, Tesla takes safety very important. It is one of the reasons why they developed Autopilot or why their cares are some of the safest, if not the safest, in crash tests. Believe Musk or not, but the statements from the company are that Autopilot is safer than when it is not in use, and this would include people who abused the system beyond its intent, so that is saying something. In fact, they are running simulations even when the car is not on Autopilot, so the company knows if Autopilot would have detected and avoided an incident instead of crashing. This isn't to say that Autopilot or Full Self Driving Mode would outperform your average driver at driving in all situations over a year because the systems still require driver engagement, but a driver with these assistant programs is undoubtedly safer than a person without them.

It is a pretty amazing thing considering where we were with cars 10 or 20 years ago:

[Irace86.2.0]

[Irace86.2.0]

[Irace86.2.0]

Originally Posted by AnalogMan

Too many people are too myopic to see it realistically for what it is, and what it isn't, and as a result people will needlessly die in the process of implementing an otherwise beneficial technology.

Does this make you feel better?

https://electrek.co/2021/03/12/tesla...-driving-beta/

Tesla removed drivers who weren’t paying attention from Full Self-Driving beta as it expands

Now Musk confirmed today that the program has been expanded to about 2,000 owners and he also said that some who were previously in the program saw their access revoked:

“FSD Beta has now been expanded to ~2000 owners & we’ve also revoked beta where drivers did not pay sufficient attention to the road. No accidents to date.”

The CEO confirmed that Tesla used its driver-facing camera to monitor driver attention in the beta program.

[Captain Snooze]

Originally Posted by Ohio Enthusiast

collective intelligence

[Captain Snooze]

Originally Posted by Irace86.2.0

because the system allows them to be complacent.

Risk compensation is a theory which suggests that people typically adjust their behaviors in response to perceived levels of risk, becoming more careful where they sense greater risk and less careful if they feel more protected.
My italics.

[AnalogMan]

NHTSB warns Tesla tests unfinished driverless tech on its users:

https://www.cnbc.com/amp/2021/03/12/...oads-ntsb.html

[Dadhawk]

Originally Posted by Irace86.2.0

Guys are we really freaking out about marketing here? Autopilot and Full Self Driving Mode are brand names. These aren’t marketing claims. .

I don't disagree with you in principle but, by the same token, if you don't believe Musk and Co intentionally used these "marketing names" to imply something the systems were not, you are mistaken. They knew exactly what they were doing when they used those names.

[Dadhawk]

Originally Posted by Irace86.2.0

Tesla could alter this behavior by pulling the car over when the driver isn't looking forward with open eyes, or it could do that if someone doesn't touch the steering wheel once a minute or something pestering, but obviously, they don't want to force someone. Maybe some states will pass laws that require such things like how some states have laws against driving while talking/texting.

To me, for Tesla to allow the car to detect a dangerous situation and ignore it puts them squarely liable when something goes wrong. Either the car requires attention when in autopilot or it doesn't. If it can detect the appropriate level of attention isn't being paid, and basically ignores it, then the company is partially at fault.

[Irace86.2.0]

Originally Posted by Dadhawk

I don't disagree with you in principle but, by the same token, if you don't believe Musk and Co intentionally used these "marketing names" to imply something the systems were not, you are mistaken. They knew exactly what they were doing when they used those names.

But those systems are Autopilot by the definition I posted, and they are FSD. FSD is in beta, so it isn’t even released to the fleet completely, which is why it is level 2 autonomy, but it is FSD 4 or 5 by that standard in terms of its capabilities. Most people don’t even know the six levels of autonomous driving. They understand the concept of FSD. Humans are capable of FSD, but everyone is different in their ability to see, speed of reaction and in how they react. Right now, the beta is a 110 year old with 20/60 vision and a Dowager's hump. By the time it is released to the fleet, it’ll be a lot better. Even when it is released, it’ll be 4 or 5 level autonomy, but Tesla will claim level 3 to keep people vigilant and cover their butts for the 0.1% of scenarios. Only when we see Robotaxis are they making a claim that the taxi is safer than a human behind the wheel.

https://www.synopsys.com/automotive/...ng-levels.html

Originally Posted by Dadhawk

To me, for Tesla to allow the car to detect a dangerous situation and ignore it puts them squarely liable when something goes wrong. Either the car requires attention when in autopilot or it doesn't. If it can detect the appropriate level of attention isn't being paid, and basically ignores it, then the company is partially at fault.

But we have states where you can drink and drive as long as you aren’t drunk, and we have states that allow you to have a phone in your hand texting or talking while driving, and we have states without tint laws. Every states determines there level of personal liberties.

Regardless, I already posted the article showing Tesla pulled FSD beta from some drivers because they weren’t paying close enough attention.

[Irace86.2.0]

Originally Posted by AnalogMan

NHTSB warns Tesla tests unfinished driverless tech on its users:

https://www.cnbc.com/amp/2021/03/12/...oads-ntsb.html

We know this. I already posted this in videos and mentioned it. Most of the beta users are Tesla employees and selected individuals. The new batch of testers expanded from 1000 to 2000, but some people weren’t engaged enough, so Tesla removed them from the beta group, which means they are heavily monitoring the drivers.

Tech companies roll out betas to developers and anyone who wants to assume the risk like with Apple releasing a beta of IOS that could possibly cause the iPhone to crash or have data loss or brick a phone.

Tesla has been running simulations of FSD for years on cars in the fleet to get the tech where it is now, but simulations only go so far. Eventually you need to test cars in real life:



Eventually, the tech graduates from company cars in controlled environments to the open roads and then to select individuals and then to the masses. Even once it is released fully, it may still disengage or revert to Autopilot if drivers aren’t focused on the road, and it will continue to rapidly improve.

[Dadhawk]

Originally Posted by Irace86.2.0

Tech companies roll out betas to developers and anyone who wants to assume the risk like with Apple releasing a beta of IOS that could possibly cause the iPhone to crash or have data loss or brick a phone.

Except those betas do not put me physically at risk because the user is willing to accept, but not adhere to, the restrictions.

As was said earlier, full autonomy will most likely be functional before it is fully released, since that will require the tech company to take responsibility for faults in the system.

I've been in tech long enough (including time as a programmer) to know that while the system ultimately be better than a human driver, it will be a long time before every contingency is programmed into the system, machine learning or not. Until it gets to 99.999% reliability the tech company would be ill advised to call it autonomous.