Tesla's Autopilot Scam - by BS Exposed

[MT]

[BrianHG]

Old news.  I claimed that it was impossible on this tech forum years ago in a Musk thread and that others will eventually beat him to it within the next 15 years (now around 12 left) and I was booed and disciplined about Musk's greatness and fantastic 2 super neuro chips were overkill and impossible to fail...

[lezginka_kabardinka]

Old news.  I claimed that it was impossible on this tech forum years ago in a Musk thread and that others will eventually beat him to it within the next 15 years (now around 12 left) and I was booed and disciplined about Musk's greatness and fantastic 2 super neuro chips were overkill and impossible to fail...

Yeah. Being shot down appears to be a recurring issue on forums full of “experts”. No one likes to admit that they don’t know everything, and could be wrong.

The opening few seconds are amusing: “it’s financially insane to buy anything other than a Tesla”

Let’s correct that:

It’s financially, but also generally insane to buy a Tesla, or ANY electric car.

Watch this space for the next 5-15 years, and you’ll see the eventual (silent) burying of EV ^anything^, it’ll be as if it was never in the public vocab, due to embarrassment and the fact it was never gonna work out, long term.

Remember 3D TVs? Nor does anyone else. I mean they do, but they’ve “forgotten”, as the hive mind has moved on.

Sci-fi fun and games. Enjoy your toys.

[Dan123456]

I do love CSS.

For anyone who hasn’t watched them before, they used to be Elon fans. But then they realised a lot of his claims weren’t adding up so they provide well researched and evidence based arguments against his BS.

As for FSD, I am still completely blown away that they are still allowed to call it “Full Self Driving”. It is such a misleading name for what it really is and am really not surprised some people over estimate its abilities.

[wraper]

For anyone who watches CSS, even though Thunderf00t has Musk derangement syndrome, is arrogant and hardly admits any mistakes, he's a scientist and overall a knowledgeable guy who is worth watching once in a while. CSS is just a stupid grifter who feeds on Musk hate exclusively and his content falls apart under slightest scrutiny. You need to keep your head deep in the ass to like and trust all that CSS says.

[Dan123456]

For anyone who watches CSS, even though Thunderf00t has Musk derangement syndrome, is arrogant and hardly admits any mistakes, he's a scientist and overall a knowledgeable guy who is worth watching once in a while. CSS is just a stupid grifter who feeds on Musk hate exclusively and his content falls apart under slightest scrutiny. You need to keep your head deep in the ass to like and trust all that CSS says.

I would agree with you RE: TF. He has some points but does waffle and repeat himself a lot. And I do agree that he really struggles to admit when he is wrong.

That said, what examples do you have for CSS’ work falling apart under scrutiny?

I’m not being a dick. I genuinely just would like to know as like to make my own informed decisions and would be happy to say I am wrong if I am 

[wraper]

That said, what examples do you have for CSS’ work falling apart under scrutiny?

I’m not being a dick. I genuinely just would like to know as like to make my own informed decisions and would be happy to say I am wrong if I am 

Here is an example of his Starlink debunking video taken apart (note it consists of 4 parts, links on the bottom) https://littlebluena.substack.com/p/common-sense-skeptic-debunking-starlink

[Dan123456]

That said, what examples do you have for CSS’ work falling apart under scrutiny?

I’m not being a dick. I genuinely just would like to know as like to make my own informed decisions and would be happy to say I am wrong if I am 

Here is an example of his Starlink debunking video taken apart (note it consists of 4 parts, links on the bottom) https://littlebluena.substack.com/p/common-sense-skeptic-debunking-starlink

Appreciate that mate 

I’ll have a read through and rewatch the video as soon as I have a chance this arvo 

Edit: Ok, watched and read through 

The article does have some valid points RE: some calculations, download speeds and comparing starlink to cable broadband. However, I wouldn’t say CSS’ work falls over because of it.

Some notes below:

42,000 sats - maths error for some of CSS’ calculations as article says 2 stages - 12,000 in stage 1, 30,000 in stage 2 and apparently stage 1 and stage 2 won’t have satellites up at the same time (so annual costs on 42,000 sats would be incorrect - although, I haven’t verified that). However, they do want to launch 42,000 sats in total + replacements for failures so a lot of CSS’ calculations stand.

Dishy - article talks about possible changes in future and the author had heard of people getting their dish replaced free of charge. Not hard facts or written policy.

On that, there are a few points in article are “I have seen” / “I have heard of X” / “in the future” / “it’s a beta” etc. arguments. I don’t think those are valid points as are speculation / anecdotal evidence (to be fair, CSS did do this too when talking about most people leaving the dish on the ground).

Article ignores the space junk issue.

Article ignores market and financial stability for Starlink.

Over all, thank you Wraper and I appreciate you flicking that through It gave me some insight into some errors and I will keep them in mind  However, I don’t think it fundamentally changes the points made in CSS’ video 

[wraper]

Also (now abandoned) CSS prediction tracker with links to CSS twitter posts (it's real bad even for a short time it got updated) https://littlebluena.substack.com/p/common-sense-skeptic-prediction-tracker

[CatalinaWOW]

I find myself in the middle.  I think there is a lot to like about electric cars.  They aren't the answer to everything, but they do have a real place in the market.  Same with self driving in cars.  The Tesla product does an amazing job.  It can't deliver enough 9s to satisfy most peoples standards for self driving capability, but it is pretty damn impressive.  Just step back for a moment and think about it.  The evidence is pretty clear that it will deliver a 99.9% chance of completing a trip safely.  I don't really know what the real number is but suspect that it has a couple more nines in it.  The general standard in risk management circles is that it needs six nines for deployment of a truly necessary function.  Don't know if the legal profession has a real definition.  And definitely self driving is optional so more than six nines would be required if that was the only criteria.  But this is an area where the current technology (human beings) doesn't deliver that level of performance.  In a purely rational world the self driving cars would only have to perform as well or better than humans, but in several threads on this forum and elsewhere it is clear that that won't be allowed here.

Bottom line, Tesla isn't good enough for prime time now, and definitely oversold their capability.  Whether that is lying or simply not thinking through how good it needs to be is unknown to me.  But I see no evidence that Tesla won't get there (maybe by adding sensors, maybe by better software, maybe by better limits on where it can be applied or something else), and no evidence of who will get there first.  And really no reason to toss them completely to the curb.  Overselling is not unique to Tesla in this area.  Other companies have had just as much hubris, but without the financial resources or market position to dig themselves as deep a hole.

[Marco]

I claimed that it was impossible on this tech forum years ago in a Musk thread and that others will eventually beat him to it within the next 15 years

No one is going to beat them to it without true AI, yet I doubt true AI will want to be a chauffeur ... it's just plain impossible for anyone, Tesla is not special. They think it's possible, the rest is just schedule slip from their perspective.

It's less a scam than a self delusion. Almost all of the media and industry keep wrongly saying self driving isn't AI hard, why would it be a scam for Elon to still believe that? Unless you're saying the entire industry is a scam which seems a bit pessimistic.

[tom66]

Old news.  I claimed that it was impossible on this tech forum years ago in a Musk thread and that others will eventually beat him to it within the next 15 years (now around 12 left) and I was booed and disciplined about Musk's greatness and fantastic 2 super neuro chips were overkill and impossible to fail...

Yeah... it's CLEARLY impossible...
https://youtu.be/watch?v=nAxHWS5i_W0

Don't get me wrong... Tesla should not have sold these systems for $10k+ each and if I had paid that kind of money for FSD I would be disappointed.  There are remaining concerns about safety and abuse of the FSD system in its beta configuration too.

But it's pretty clear that Tesla are quite far ahead on FSD and that provided they can continue to iterate on safety and reliability, they can fundamentally achieve driverless operation at SAE Level 4 at least. (They are at SAE Level 3 currently.)

The only other companies that have achieved this are Waymo and Cruise who both require custom cars covered in LIDAR at a cost of $5-10k per sensor.  Waymo have certainly beaten Tesla to the FSD crown, no doubt, but it's not as if the technology is one and done.

[nctnico]

Old news.  I claimed that it was impossible on this tech forum years ago in a Musk thread and that others will eventually beat him to it within the next 15 years (now around 12 left) and I was booed and disciplined about Musk's greatness and fantastic 2 super neuro chips were overkill and impossible to fail...

By the way things look nowadays, it might be sooner than your 12 years. IIRC BMW and Mercedes are quite far along and operate at a higher readiness level compared to Tesla. The car electronics giant Bosch and their huge capacity to invest in new technology has a lot to do with it. It is way for car manufacturers to share development costs so each only pays a fraction of the bill (mostly afterwards). In the end Tesla will simply buy the self driving system from Bosch.

[janoc]

But it's pretty clear that Tesla are quite far ahead on FSD and that provided they can continue to iterate on safety and reliability, they can fundamentally achieve driverless operation at SAE Level 4 at least. (They are at SAE Level 3 currently.)

Tesla's Autopilot is at SAE level 2, despite the marketing and the "Full self driving" moniker. It is by their own description to the regulator (unlike their marketing message):

https://www.thedrive.com/tech/39647/tesla-admits-current-full-self-driving-beta-will-always-be-a-level-2-system-emails

The only system that is currently certified as level 3 is actually from Mercedes.

Tesla's technology is very good - but is still very very far from anything resembling a level 4 and even less a level 5 system. And it is not at all clear whether they can actually "fundamentally achieve driverless operation at SAE Level 4 at least." (I wonder where do you get that sense of security to make such claim - based on what, exactly?)

The biggest problem is that they are iterating on safety and reliability at the expense of safety of everyone else around their cars. Sure, it is the driver who is ultimately responsible for whatever the car does. But that is just legal CYA and won't help the victim of an accident - the same as a shooting victim isn't helped by the fact that the parent of the shooter should have secured their guns at home better.

Many things that Tesla does with their Autopilot are going completely against the existing human factors research and giving the drivers false sense of security, thus promoting complacency and drivers not paying attention to the road. Which is lethal in a level 2 (or even level 3) system. These lessons have been learned and paid for in blood in aviation and railways, for example, where such systems exist since a long time ago - and where the available reaction times when the automation conks out are typically much longer than in a car driving on a motorway with a distracted driver.

[wraper]

But it's pretty clear that Tesla are quite far ahead on FSD and that provided they can continue to iterate on safety and reliability, they can fundamentally achieve driverless operation at SAE Level 4 at least. (They are at SAE Level 3 currently.)

Tesla's Autopilot is at SAE level 2, despite the marketing.

Yes, it's level 2 technically. However SAE level tells very little about actual capability and more about how much driver attention is needed. You can do pretty useless autopilot with no requirement to keep your hands on steering wheel but which only works on some pre-mapped straight highways, with 40 mph speed limit, only in perfect weather, that cannot change lanes, and call it SAE level 3. So it's only useable in traffic jams on those certain perfect situations, hi Mercedes.

[janoc]

Yes, it's level 2 technically. However SAE level tells very little about actual capability and more about how much driver attention is needed. You can do pretty useless autopilot with no requirement to keep your hands on steering wheel but which only works on some pre-mapped straight highways, with 40 mph speed limit, only in perfect weather, and call it SAE level 3. So it's only useable in traffic jams on those certain perfect situations, hi Mercedes.

Well, yeah - but the issue is not the "capability" but the amount of attention it needs. And thus e.g. the level of complacency it induces in the driver and thus impact on safety and hazard to everyone around them. A very capable system that works 90% of the time is arguably worse than a less capable one that works only in 10% of situations.

That's why Mercedes has it designed in that way - they know their system isn't perfect so are erring on the conservative side to not get their backside handed to them by the regulator/courts. Compare that approach with Tesla's that does very little to prevent misuse (they have actually just issued a recall for over 2 million of vehicles over this) - and then it causes accidents because, surprise, people do use it where it shouldn't have been used, despite the fine print in the manual telling them not to.

Ultimately these level 2-3 systems are dangerous because the drivers will rely on them, become distracted/inattentive - and when their intervention is ultimately required they will not react in time - getting "back in the game" when you were not paying attention can take even several seconds that you likely don't have when at high speed.

[nctnico]

Last week Tesla has been forced to do a recall to address several safety issues:

https://edition.cnn.com/2023/12/13/tech/tesla-recall-autopilot/index.html

[MT]

A German court ruled that Tesla cannot talk about 'full potential for autonomous driving' or 'Autopilot' in its ads in the country.
https://www.caranddriver.com/news/a33338288/germany-tesla-autonomous-driving-court-ruling/

Still Tesla German owners manual 2023 talks about auto pilot:
https://www.tesla.com/ownersmanual/modely/de_de/GUID-101D1BF5-52D2-469A-A57D-E7230BBEE94B.html

Tesla appealed but was overruled 2022:
https://teslamag.de/news/exklusiv-web-werbung-tesla-autopilot-fsd-option-deutschland-zulaessig-51722

Tesla forced to buyback cars:
https://www.autoevolution.com/news/autopilot-makes-tesla-buy-back-another-ev-in-germany-it-s-the-third-one-until-now-195163.html

A pilot are not requested by Boeing/Airbus to "always hold their hands on the steering" when autopilot is on. It is well known a jet these days can take off, fly and land all by it self with no intervention of a hominid or monkey.

[Marco]

By the way things look nowadays, it might be sooner than your 12 years. IIRC BMW and Mercedes are quite far along and operate at a higher readiness level compared to Tesla.

The current level 3 systems are just in lane follow the duckling systems for slow dense traffic on highways, that's all they will ever be.

Waymo&co need RC every 5 minutes to not gridlock a city. Level 3 follow the duckling systems is the best you can do on highways with full autonomy. True self driving is AI hard.

[Marco]

Yes, it's level 2 technically. However SAE level tells very little about actual capability

It tells you about its consistent capabilities. Level 2 is fundamentally not reliable to a meaningful degree.

[BrianHG]

Old news.  I claimed that it was impossible on this tech forum years ago in a Musk thread and that others will eventually beat him to it within the next 15 years (now around 12 left) and I was booed and disciplined about Musk's greatness and fantastic 2 super neuro chips were overkill and impossible to fail...

Yeah... it's CLEARLY impossible...
https://youtu.be/watch?v=nAxHWS5i_W0

   Yup, after seeing that, another 10 years away.   Don't get me wrong, improvements are being made, but place that car in downtown New York, with hundreds of pedestrians at each corner, plus cyclists and motorbikes swerving, squeezing between cars, dogs running across the road... with all that steering hesitation and braking at the slightest question...  Streets lined with double parked cabs & buses making aggressive moves into lanes, current FSD is nowhere near ready nor does it yet have more than the initial image & environment recognition.  It's missing the understanding and intent recognition of all the other actors it sees on the road.  But that level of recognition will come and what you showed me in that video will be looked upon like us being fools to trust such simplistic dangerous garbage in a decade when we will see the first authentic FSD cars which will come with no steering wheel or gas/accelerator pedals installed.

[CatalinaWOW]

Just got around to actually watching the video and find it as far from the truth as Musk is, but in opposite direction.

I don't own a Tesla, but do have a few acquaintances who do, and my son has traveled in one for many hours.

1.  The constant errors and jerkiness shown in the video clips have not been observed by them.   This doesn't mean that they don't happen, but it does mean that the frequency of occurrence is much lower than implied in the video.

2.  The video reports with horror the reports of people asleep at the wheel and video of someone climbing into the back seat.  I agree that this is stupid, and not warranted by the current performance, but again the fact that this is happening fairly often indicates that the incidence of crashes is fairly low, not what this video claims 100% chance of a crash in a short time.

3.  Claims of scamming and widespread customer dissatisfaction seem overblown also.  Those I know who have the FSD are completely happy with it.  Ignorance may be bliss, but they are generally well educated and informed people (something that has a fair amount of correlation with being able to afford a Tesla).

[coppice]

Old news.  I claimed that it was impossible on this tech forum years ago in a Musk thread and that others will eventually beat him to it within the next 15 years (now around 12 left) and I was booed and disciplined about Musk's greatness and fantastic 2 super neuro chips were overkill and impossible to fail...

Yeah... it's CLEARLY impossible...
https://youtu.be/watch?v=nAxHWS5i_W0

   Yup, after seeing that, another 10 years away.   Don't get me wrong, improvements are being made, but place that car in downtown New York, with hundreds of pedestrians at each corner, plus cyclists and motorbikes swerving, squeezing between cars, dogs running across the road... with all that steering hesitation and braking at the slightest question...  Streets lined with double parked cabs & buses making aggressive moves into lanes, current FSD is nowhere near ready nor does it yet have more than the initial image & environment recognition.  It's missing the understanding and intent recognition of all the other actors it sees on the road.  But that level of recognition will come and what you showed me in that video will be looked upon like us being fools to trust such simplistic dangerous garbage in a decade when we will see the first authentic FSD cars which will come with no steering wheel or gas/accelerator pedals installed.

Getting rid of the steering wheel means the car needs to be able to do all the little manoeuvres needed to put the car exactly where you want it when you park. I don't think they've even begun to address that kind of thing. "Oh, its really sunny. Lets leave it in the shade of that tree." I don't think is a part of the current design.

A lot of the effective flow at busy junctions only works when you smile at another driver and beckon them to go. If a self-driving car waits for a gap to pull into if could be waiting through the entire rush hour period at many junctions. 

[MT]

I find myself in the middle.  I think there is a lot to like about electric cars.  They aren't the answer to everything, but they do have a real place in the market.  Same with self driving in cars.  The Tesla product does an amazing job.  It can't deliver enough 9s to satisfy most peoples standards for self driving capability, but it is pretty damn impressive.  Just step back for a moment and think about it.  The evidence is pretty clear that it will deliver a 99.9% chance of completing a trip safely.  I don't really know what the real number is but suspect that it has a couple more nines in it.  The general standard in risk management circles is that it needs six nines for deployment of a truly necessary function.  Don't know if the legal profession has a real definition.  And definitely self driving is optional so more than six nines would be required if that was the only criteria.  But this is an area where the current technology (human beings) doesn't deliver that level of performance.  In a purely rational world the self driving cars would only have to perform as well or better than humans, but in several threads on this forum and elsewhere it is clear that that won't be allowed here.

Bottom line, Tesla isn't good enough for prime time now, and definitely oversold their capability.  Whether that is lying or simply not thinking through how good it needs to be is unknown to me.  But I see no evidence that Tesla won't get there (maybe by adding sensors, maybe by better software, maybe by better limits on where it can be applied or something else), and no evidence of who will get there first.  And really no reason to toss them completely to the curb.  Overselling is not unique to Tesla in this area.  Other companies have had just as much hubris, but without the financial resources or market position to dig themselves as deep a hole.

Overselling is by law called for what it is, a scam, i.e scam'ed into paying for something that is not 100% there.

Andrej Karpathy, who formerly was head of AI for Tesla’s Autopilot and FSD products, outlined their reasoning behind removing both the radar and ultrasonics from Tesla cars, as well as never using LIDAR or maps. While Elon Musk is best known for making statements on this, Karpathy was his go-to guy on backing up that reasoning. Karpathy raised eyebrows, however, when earlier this year he took a sabbatical from the job and eventually announced he would leave it.

https://www.forbes.com/sites/bradtempleton/2022/10/31/former-head-of-tesla-ai-explains-why-theyve-removed-sensors-others-differ/

[wraper]

Andrej Karpathy, who formerly was head of AI for Tesla’s Autopilot and FSD products, outlined their reasoning behind removing both the radar and ultrasonics from Tesla cars, as well as never using LIDAR or maps. While Elon Musk is best known for making statements on this, Karpathy was his go-to guy on backing up that reasoning. Karpathy raised eyebrows, however, when earlier this year he took a sabbatical from the job and eventually announced he would leave it.

And what are you implying here? Here is what Andrej said 1:44:28 - Leaving Tesla. Also he said he may return to Tesla in future.

“I think over time during those five years; I’ve kind of gotten myself into a little bit of a managerial position. Most of my days were, you know, meetings and growing the organization and making decisions about sort of high-level strategic decisions about the team and what it should be working on and so on.”

“And it’s kind of like a corporate executive role, and I can do it. I think I’m okay at it, but it’s not like fundamentally what I enjoy, and so I think when I joined, there was no computer vision team because Tesla was just going from the transition of using Mobileye, a third-party vendor, for all of its computer vision to having to build its computer vision system. So when I showed up, there were two people training deep neural networks.”

[coppice]

Andrej Karpathy, who formerly was head of AI for Tesla’s Autopilot and FSD products, outlined their reasoning behind removing both the radar and ultrasonics from Tesla cars, as well as never using LIDAR or maps. While Elon Musk is best known for making statements on this, Karpathy was his go-to guy on backing up that reasoning. Karpathy raised eyebrows, however, when earlier this year he took a sabbatical from the job and eventually announced he would leave it.

And what are you implying here? Here is what Andrej said 1:44:28 - Leaving Tesla. Also he said he may return to Tesla in future.

“I think over time during those five years; I’ve kind of gotten myself into a little bit of a managerial position. Most of my days were, you know, meetings and growing the organization and making decisions about sort of high-level strategic decisions about the team and what it should be working on and so on.”

“And it’s kind of like a corporate executive role, and I can do it. I think I’m okay at it, but it’s not like fundamentally what I enjoy, and so I think when I joined, there was no computer vision team because Tesla was just going from the transition of using Mobileye, a third-party vendor, for all of its computer vision to having to build its computer vision system. So when I showed up, there were two people training deep neural networks.”

What did you expect him to say? "I made some dumb decisions, and decided to get out while the going was good"? Over the last few years each time Musk has made a bold prediction about FSD, a bunch of the senior people have quickly left. That's not a good sign.

[wraper]

What did you expect him to say? "I made some dumb decisions, and decided to get out while the going was good"? Over the last few years each time Musk has made a bold prediction about FSD, a bunch of the senior people have quickly left. That's not a good sign.

Huh? If he wanted similar position, he could have landed at different company the next day. But he did not.

[coppice]

What did you expect him to say? "I made some dumb decisions, and decided to get out while the going was good"? Over the last few years each time Musk has made a bold prediction about FSD, a bunch of the senior people have quickly left. That's not a good sign.

Huh? If he wanted similar position, he could have landed at different company the next day. But he did not.

If he has some suitable share options he might be easily able to relax for a while. Not everyone is desperate to get their next job, especially if they have just left a high stress environment, which a lot of people imply Tesla is.